Merge 3449-3535 from the NKI branch containing:
I Bug fixing concerning:
1) Value "NS" for cellHTS2 parameter "Normalize Plates NegControls" was forwarded to cellHTS2 as "N" in stead of "NS"
2) Changed "NS" into "S", because when giving the value "N" for the parameter "Normalize Plates NegControls", cellHTS2 will take both "N" en "NS" wells for calculating the median.

II Added columns time_point_ordinal and zdepth_ordinal to table result_value_type, for storing information from high content microscope.

- Updated example database, applying all required migration scripts
- Updated 3431.sql, 3440.sql to remove (more) ICCB-L specific updates, and made fixes

Updated displayed version number for public release

Updated for changelog.html, newsViewer.xhtml, readme for public release

removed (recent) ICCB-L specific messages in login{,-error}.xhtml

updated SQL migration scripts to remove or identify ICCB-L specific changes

Merged library-mgmt-rework branch (covers ICCB-L v1.7.4, v1.7.5 releases, originally developed on /branches/iccbl/trunk, and v1.8.0 release, developed on /branches/iccbl/ss-library-mgmt-rework) to main trunk: merge -r3018:HEAD http://forge.abcd.harvard.edu/svn/screensaver/branches/iccbl/library-mgmt-rework.  See changelog.html for details.

Merging in some older, ICCB-L-specific schema migration scripts.  These were previously not merged in since they were ICCB-L-specific, but have decided to merge now, since many of the other migration scripts are also ICCB-L-specific, and it will reduce confusion to not have different sets of scripts on different branches.  In particular, now that ICCB-L intends to keep in sync with /trunk for its development and production releases.  We'll find a more general way to deal with facility-specific migration scripts.

Added to the sql-script for adding column owner, an insert statement in schema_history.

Now owner can be added to a library (optional) and/or be edited.

Adding field owner to library, to enable hiding validation libraries for other people in screensaver. (Code to fill the field via UI instead of sql will follow).

fix compilation errors that appear when using Eclipse Galileo (3.5)

Added the following options to run cellHTS2
1 Normalization by
a) (median) negative controles.
b)  loess.
c)  NPI

2.NormalizePlatesScale: Additive vs Multiplicative.
 In case of normalization by mean/median (of samples) or by (median) of negative controles, additive means subtracting the mean/median where as multiplicative means dividing by the mean/median.

3. Selection of type of negative control: N vs NS

A change to the R cellHTSDb library was made, so you need to install the latest version (=0.3) .

At NKI we now run with cellHTS2 version 2.8.0 and R 2.7.0.

[#1735] CellHTS2 should handle sparse ResultValues on plates'
Added JUnit test for screenResult with no Resultvalue for well AO1.
Solved problem by initializing the double[][] with CellHTS2.NA values. 

added commented-out, alternate versions of cellHTS commands that I had used in the past to produce analysis results that matched the Boutros tutorial; functionally unchanged.

merge -r2962:HEAD http://forge.abcd.harvard.edu/svn/screensaver/branches/iccbl/trunk

added warning for changing default plate size selection

Merged plate-size-rework branch back into main trunk, using:
   merge --ignore-ancestry http://forge.abcd.harvard.edu/svn/screensaver/trunk@2979 http://forge.abcd.harvard.edu/svn/screensaver/branches/plate-size-rework@HEAD .
(These non-standard merge options were necessary, since the branch was not created as a clean copy of a trunk revision, but with edits already having been made.)
To see diffs merged in:
   diff -r 2979 http://forge.abcd.harvard.edu/svn/screensaver/trunk http://forge.abcd.harvard.edu/svn/screensaver/branches/plate-size-rework

Merge from NKI branch rev 3006:3007. Containing:
In the cellHTS2 report the following elements were not generated:
1. heatmaps for intensities per plate and replicate
2. Screen-wide image plot of the scored values
Ad 1: Due to the setting of the write report argument plotPlateArgs = FALSE
Change it to TRUE
Ad 2: Due to lacking of the "scored" element in the cellHTS2 parameter of the writeReport.
Have added it.
I also included changes to the R and JUnit test voor writeReport. 

Merge rev 2070 from collaboration branch:
Made running the cellHTS2 writerReport function configurable for the parameter progressReport. WriteReport runs default with a progress window which requires the user to have access to the display.  Running it with progressReport= FALSE, no progress window is shown, which on a server is obsolete anyway. Now possible to run Rserve under account without access rights to display. Created new property in the screensaver.properties file: cellHTS2report.writeReport.progressReport=FALSE. Alternatively set to TRUE.
See for example cfg/screensaver.properties.web

Merged from collaboration branch
rev. 2963
CellHTS2 output can now be written to location outside the webapps directory. The root directory for the output is now in stead of webapps, defined in the property cellHTS2report.filepath.base in the screensaver.properties. See example cfg/bii/screensaver.properties.bii.sample. The data retrieval is done by a new class DataServlet in the cellHTS2 package. The writing and reading is working, however the presenting of the url on the screen isnot yet. 

rev. 2964
In ScreenResultviewer the check if file exists is now correctly working in combination with the feature of full path for cellhts output directory.

fix for unit tests

 [#1716] UI Screen Delete Option controlled by properties file setting
 This test, while not necessary for the UI Delete Option, was created to provide some test coverage for the backing business logic.  Also, the test was updated to separate the create and delete steps into transactions.

 [#1716] UI Screen Delete Option controlled by properties file setting
 this set of files does not belong with this feature commit.

 [#1716] UI Screen Delete Option controlled by properties file setting
 Also, includes a refactoring of the delete feature into the AbstractEditableBackingBean

minor refactorings for property that configure use of tree-based versus list-based data table column selection UI.

fixed web.xml, for recently moved classes

fixed spring-context-ui.xml, for recently moved classes

Moved WebDataAccessPolicy into e.h.m.iccbl.screensaver.policy, since this class represents an ICCB-L-specific policy.

Removed sg.edu.astar.bii package, moving all of its classes under edu.harvard.med.screensaver, as all of these classes are general purpose and not BII-specific implementations.  IMCB/BII had approved of this a while back.  Credited authors with @author tag in (non-trivial) classes.

- updated generic screensaver.properties.* files with cellHTS option and genericized the feedback email address
- removed BII's screensaver.properties, since goal is to have no deployment-specific files 

minor fixes to web.xml

Removed erroneously merged code for #1470 ("hide small molecule options"), since we're not yet accepting this contribution on the trunk.

 [#1471] reorder data table's visible columns
 Integrate the IMCB code for reodering columns as a new facelet widget "columnSelectorReorderList" with the searchtable.  This will be a toggled feature for now, to use, set the value in the screensaver.properties file:
searchtable.use_reorder_list_widget=true
setting this property to true will remove the alternate widget, the "columnSelectorTree" widget, or the stock screensaver tree selector, from all search tables.

Merged [#1473] (display assay well type in screen result table) from branches/imcb/screensaver-1.7.0-share to /trunk.

merged in cellHTS UI in screenResultViewer, which did not make it in during the previous large merge (many of the UI files for screenResultViewer were refactored).

- Merged from iccbl/trunk, taking changes from ICCB-L versions 1.7.1, 1.7.2, and 1.7.3
- svn merge -r 2726:2958 http://forge.abcd.harvard.edu/svn/screensaver/branches/iccbl/trunk
- Compiles, but NOT yet tested.  Intention is to tag a future revision with tag upon official release.
- See changelog.html for all tickets addressed.  
- For full change log from commit log use 'svn diff -r 2726:2958 http://forge.abcd.harvard.edu/svn/screensaver/branches/iccbl/trunk'

- merge -r2736:HEAD(2775) http://forge.abcd.harvard.edu/svn/screensaver/branches/collaboration http://forge.abcd.harvard.edu/svn/screensaver/trunk
  - [#839] 96-well plates
  - [#837] multi-channel

a sample screensaver.properties files

a sample screensaver.properties files

- added section headers to cellHTS2Runner.xhtml as a cosmetic improvement
- removed unused code in ScreenResultViewer

To associate with new tracker item [#882]. The old RT ticket #159853 is moved to the ABCD Forge tracking system #882. 

(Changes from collaboration branch)
Servlet filter for cellHTS2 report authorization [#159853]
 - This filter checks whether the logged-in user has the rights to view the screenresult before he/she can view the cellHTS2 report. If the user is restricted from viewing, he/she will be redirected to the cellHTS2ReportAccessDenied page.
 - Similarly for cases where the ScreenViewer has not been visited (refer to remark below), the user will be redirected to the same page as above.
 
Remark from Andrew:
 It checks the access restrictions on the screen being currently viewed (as of the previous HTTP request, meaning that ScreenViewer must have already been visited by the user)

Originally added to the collaboration branch.

merge http://forge.abcd.harvard.edu/svn/screensaver/trunk@2730 http://forge.abcd.harvard.edu/svn/screensaver/branches/collaboration@2736 

part of merge

Updated Eclipse .classpath project file to use upgraded postgresql jdbc driver jar.

Resolved [#835] (hibernate fails on LOB IDs > Integer.MAX_VALUE): upgraded postgresql jdbc jar

- Resolved [#834] (entity viewer edit/save button positions): duplicated buttons at bottom of entity viewer edit pages
- In entityViewerDetails.xhtml:
   - Moved special ID# and Created Date fields from upper-left to properties list. 
   - Added hideId param to entityViewerDetails page to optionally hide when a semantic ID is available (e.g. screen number)

Resolved [#833]

updated overview.html text (mostly to test new ABCD Forge repository access)

- Resolved #153821: screen viewer load performance.  Aside from fixing #151940
(see below), improved performance in the case where a screen is opened via the
ScreenFinder, which has to do a full ScreenSearchResult initialization prior
to finding and viewing the specified screen.  To reduce the time for this
search result initialization, the N+1 select behaviors for
Screen.billingInformation and Screen.labHead.labAffiliation have been
eliminated.

    - BillingInformation and BillingItem are no longer entities, and are
    instead value types.  The BillingInformation entity has therefore been
    folded into Screen, at the physical schema layer.  This avoids the N+1
    select problem between Screen and BillingInformation, as
    BillingInformation is now always loaded along with the Screen, and a
    Screen now always has associated BillingInformation, even if its
    properties are all unused.  The BillingItem collection is now
    (temporarily) associated with Screen, rather than BillingInformation,
    since HqlBuilder does not yet support queries for collections hanging off
    of an embeddable type.

    - ScreenSearchResults eager fetches LabHead.labAffiliation, addressing the
    N+1 select problem.

- Resolved #151940: screens with many cross-screen data headers load slowly

    - Columns for ScreenResultViewer data table are calcuated lazily, when
    Data section is expanded by user. Also, when scrolling through screens,
    Data and Heat Map sections are automatically closed.  Together, these
    reduce the initial load time for viewing a new screen.

    - WellSearchResuls avoids loading all RVTs and AnnotationTypes when opened
    in a mode for which the primary entity type is null (e.g., when used to
    view well data for a null ScreenResult, null Library, etc.).  It was
    entirely unnecessary to have been loading this data, and only reduced
    performance.

Resolved #153036: user viewer: hide 'add screen' buttons for screeners

LibrarySearchResults exposes MIRNA_{INHIBITOR,MIMIC) library types.

Reverting part of previous commit.  Decided to remove the independent web pages for Lab Members Browser, Screen Associates Browser, and Associates Browsers.  Since UserViewer was not aware of which browser was "current", it always attempted to show users in the Screeners Browser.  Properly handling determination of the "current" browser was going to be too involved, so reverting back to having a single web page and backing bean for screener search results.

- Resolved #155077: screener search fails after browse lab members
- Screener search is always reinitialized prior to performing search, ensuring that additions and changes to users are considered by the search. 
- Created independent web pages for Lab Members Browser, Screen Associates Browser, and Associates Browser, which share a separate search result backing bean from the one used by Screeners Browser.  This was not technically necessary to resolve the above bug, since the Screener search is always reinitialized.  But it's arguably a better design, since a future optimizations might have us only fully reinitializing the Screener Search Result in response to user change events.

Resolved #153997: user viewer 'screens' panel open by default

Resolved #155068: users browser default sort desc on ID

- Resolved #155060: miRNA library types.  Added 'miRNA Inhibitor' and 'miRNA Mimic' library types.
- Updated javadoc for Library

javadoc warning/error fixes

updates to example-database/ files

Removed ICCB-L specific constraint regarding CherryPickRequest numbers.

factored out a 'compile' target from the 'build' target in build.xml; this allows the 'ddl' target to run w/o specifying the screensaver.properties.file option.

build.xml updates/fixes

fixed syntax error in build.xml

minor build.xml enhancements:
  - build target fails if screensaver.properties.file is not defined (a common oversight)
  - stop target does not use catalina.sh -force flag, since it shows an error, and normal shutdown is working fine w/Tomcat 5.5
  - default target is 'build', which is less destructive than 'deploy'

cfg/ file update

updated initialize_example_users.sql

Tomcat 5.5 is now supported, and Tomcat 5.0.x is no longer officially supported.  Mostly required minor changes to the one-time Tomcat configuration file, screensaver.xml.

updated javadocs

Updates to database/ contents

updated README.txt 

updated javadocs

update web app meta data

Turned off Facelets 'development' flag, as the default setting for production deployments.

updated javadocs

fixed build.xml (broke due to earlier move of web/tomcat dir to cfg/)

Removed ICCB-L specific command-line applications: IccbCompoundsStudyCreator, MedicinalCompoundsStudyCreator, ICBG report generator.  Moved to screensaver-iccb/ svn directory.

updated main README.txt

Removed a one-time command-line app used at ICCB-L.

- Moved web/tomcat to cfg/tomcat, since the member files are for a one-time Tomcat configuration step, rather than ongoing deployment
- Removed web/tomcat/server.xml since his is an ICCB-L specific file (moved to screensaver-iccbl/ top-level svn directory)

run.sh no longer needs to care about 'unused/' lib directories

removed ICCB-L specific bin/ scripts (moved to screensaver-iccbl/ svn dir)

added comment

- Restructured lib/ directory, creating category directories. Removed unused/ subdirectories and member jars.  Renamed dev/ directories to .dev/.  
- Moving servlet-api-2.4jar to web/.dev/ prevents deployment of this jar, avoiding Tomcat warning during web app startup

updated a README.txt file

remove unused developer-specific Ant target

- Updated .classpath file to contain latest jar dependencies, without any 'sourcepath' attributes (which are developer-specific)
- Updated README.txt files

Collapsed developer-specific Eclipse JDT settings into a single file, which contains the suggested Java development settings.

Removed user/host-specific directories under cfg/.  Instead, this directory now contains a single set of prototypical configuration files.

- Removed flotsam+jetsam/ directory, since this was mostly specific to ICCB-L.  Moved to a separate subversion top-level directory, 'iccbl'
- Added database/ directory for schema migration scripts and example database files, previously located under flotsam+jetsam/

Revoved .eclipse.prefs/ as this was not being maintained and likely unusable with current versions of Eclipse.

update to personal log4j.properties.debug file

Improved robustness of WellDeprecator file parser

Removed ScreenDBSynchronizer!  No longer used now that Screensaver 1.7.0 has fully replaced ScreenDB.

fixed lab_affiliation_categories.txt delimiter

updated news for 1.7.0

updated text for null value of lab head in user viewer edit mode

updated version to 1.7.0

added lab affiliation

updated a hack for a special well viewer library message, due to library name renaming (hacks will always come back to haunt you!)

fixed unit test

UI/css tweaks & fixes

Minor fix to UserSynchronizer to permit lab heads that have themselves listed as a lab head in ScreenDB; log output fixes

ScreensaverUserRole.SCREENER is hidden in User Viewer UI

added log output for troubleshooting data users by UserSynchronizer

- Resolved #151929: checklist item changes

  - Updated checklist item set: removed, added, renamed various items, per requirements

  - Added ChecklistItemGroup enum and corresponding 'group' property to ChecklistItem. UI uses this for visual partitioning of checklist items.

  - Added a Legacy ChecklistItemGroup, allowing database to contain various checklist items that are hidden from view in the UI

  - Activated/Completed checklist item statuses shown in 'eyeCatcher' style

- Resolved #151939: user viewer: revise field ordering
 
  - Moved UserViewer's labName and labAffiliation fields towards top

  - Moved UserViewer's roles table to the left side

- Partial resolution of #151964: user viewer: emphasize primary fields.  

  - Made UserViewer's user name and lab fields "primary", which affects only their visual emphasis.  Visual effect not working for lab name.
  
  - property.xhtml has new 'primary' attribute.

- Updated some css styles for fixed column widths

- ScreenDBSynchronizer now catches erroneous case when a lab head has a lab
head himself.

'Dropped*' status values are (once again) mutually exclusive with the Completed and Never Initiated.

fix for WellVolumeSearchResults instantiation in spring-context-ui.xml

fix to menu.xhtml for IE, reinstating buttons for search forms; cosmetic updates

Updated text in main.xhtml, pointing out new features.  Welcome message hidden for administrators.

- Updated newsViewer.xhtml for 1.6.3
- Search result tables use dataText style for table content

- Resolved #151396: user search on menu panel

- Minor reorganization of menu panel

- Fix made to EntitySearchResults.buildRowPageSelectListener(), to behave properly if called before the search result has been rendered.

- Logged in user name in menu panel is a now a link to UserViewer, for viewing self

- Fixed some access rights in userViewerBody.xhtml.

- User can view his/her own ID fields in UserViewer.

- Resolved #151579 (browse collaborators result set is wrong).  

- 'My Collaborators' menu command is now 'My Associates', and opens a UserSearchResult
with the screener's lab head, fellow lab members and all screen associates (collabs, 
lab head, lead screener), as determined by SRU.getAssociatedUsers().

- WebDataAccessPolicy now restricts ScreeningRoomUser to "associated" users,
as determined by SRU.getAssociatedUsers().

- Added ScreenSearchResults.searchAssociatedUsers(), to provide an explicit mechanism 
for browsing this set of users.  Previously, we relied upon the data access policy to
simply restrict the search result to only associated users, but this is was both an 
inefficient and obtuse design.

- Fixed ScreeningRoomUser.getAssociatedUsers(), which was only returning
collaborators (was unused previously)

- Fixed bug in UserViewer that caused exception for ScreeningRoomUser login
accounts.

Fixed bug whereby ActivityViewer did not show the correct performedBy user when activity type was a CherryPickLiquidTransfer.  The set of performedBy selections needed to include all ScreensaverUsers, not just ScreeningRoomUsers; the selections are limited to only ScreeningRoomUsers when activity type is a Screening activity, rather than a LabActivity.

- Resolved #151401: data table search does not update entity viewer

- The fix required overriding the filter() method of the DataTableModel returned
by EntitySearchResult, in order to catch the invocation of the filter
operation by DataTableModelLazyUpdateDecorator (rather than overriding the
EntitySearchResult.refilter() operation, which would cause the hook to execute
prematurely and too often).  Re-implemented the EntitySearchResult.resort()
override in the same way, for design consistency.

- Removed VirtualPagingDataModel.notifyListenersOfRowIndexChange(), as this
was no longer being used (even before above changes).

- Introduced DataTableModelType enum and DataTableModel.getModelType(), in
order to avoid having to rely upon checking the class type, which is often an
"unexpected" subclass or decorator class, introduced by various levels of
indirection.

- Fixed some DataTable/SearchResult unit tests.  Updated unit test
PackageTestSuite hierarchy, which had ignored (and thus failed to invoke) many
unit test classes.

merge -r2487:HEAD(2586) /gf/project/screensaver/scmsvn/?action=browse&path=svn%2Bssh%3A%2F%2Fant4@orchestra.med.harvard.edu/svn/iccb/screensaver/branches/release-1.6.1-maint

Fixed LazyInitEx in ScreenViewer

- Fixed web.xml permissions problem with staffBrowser.xhtml by moving to non-admin directory, since readEverythingAdmin role should have access to staffBrowser, and not just usersAdmin role.
- Added temporary check and error message if UserViewer is attempted to be open to view an AdministratorUser (staff).

fixed screenResultViewer.xhtml, which was not showing heat map viewer

Tentative resolution for #150528: roundedDiv display problems of firefox 3.  Encapsulated all s:roundedDiv UI components into ss:roundedDiv, which now no longer uses t:roundedDiv, but CSS styles.  Rounded corners only supported in Mozilla and CSS3 supported browsers.  Works on FF2 and FF3; not yet tested on IE{6,7}, but should hopefully work with non-rounded corners w/o other display issues.

fixed unit tests

Modified StatusValue ranks for "Dropped*" and Transferred*" values to be independent from "Completed*" and "Never Initiated", to accommodate legacy data, which does not hold these values to be mutually exclusive, and this seems reasonable enough to allow a screen to be transferred/dropped even after being completed or not initiated.

Fixed UserSynchronizer to only add comment to ChecklistItemEvent if non-empty

- Updates and fixes to ScreenSynchronizer and UserSynchronizer to accommodate data model changes since 1.6.1 (last ICCB-L production release), including LabHeads, ChecklistItems, StatusItems.  
- Updated 2540.sql to include optional SQL statements to entirely clear the LabHead association on Screen and ScreeningRoomUser, in order to bring existing production data into compliance with data model constraints; ScreenDBSynchronizer recreates these associations in a compliant manner.
- UserSynchronizer recreates all ChecklistItems in the database, per ICCB requirements for updated set of ChecklistItems.

- ActivityViewer edit mode takes place in independent viewer page, outside of search result
- Browsing lab activities from ScreenViewer initialized LabActivitiesBrowser with just the lab activities from the screen

- Work towards #146044: entity viewers always within search results context
- ActivityViewer is now always shown within the context of a ActivitySearchResults

activityViewerBody.xhtml now uses the entityViewerDetail.xhtml template, for consistent layout

screenDetailViewerBody.xhtml now uses the entityViewerDetail.xhtml template, for consistent layout

- Partial resolution of #147245: screen viewer cosmetic ehancements
- Admin-only fields shown in maroon foreground text color
- Added 'adminOnly' attributes to property{,Group}.xhtml

UserViewer screen tables are expanded by default

UserSynchronizer now creates ChecklistItem entities, which has been seeded with the official list, per requirements/feedback.

- Corrected column header in checklist items table in user viewer
- Added popup.xhtml and textColumn.xhtml UI facelet components

- Layout improvements to UserViewer

- Added entityViewerDetails.xhtml facelets template, which will provide a more consistent implementation of entity viewers.  Only used by userViewerBody.xhtml, currently.

- UserViewer checklist item tables now allows the 'performed by' date to be
edited (but only once, at creation time, to guarantee model constraints).

- Added data model constraint checks to ensure that ChecklistItemEvents are
being activated and expired in the approriate order.

- ScreeningRoomUser.checklistItemEvents is now a SortedSet, since the ordering
of activation/expiration events is significant

- Added ScreeningRoomUser.getChecklistItemEvents(ChecklistItem), to retrieve
events of a particular type, and in order

fixed a unit test

- Resolved #147993: user checklist items data model changes:

   -  ChecklistItem is now named ChecklistItemEvent.  

   -  ChecklistItemType is now named ChecklistItem.

   - A ScreeningRoomUser can now have multiple ChecklistItems, one per
   activation/expiration/completed event.  This allows "expirable" checklist
   items to be repeatedly activated and deactivated.

   -  ChecklistItem now has an associated AdministrativeActivity

- Resolved #147995: user viewer: checklist item table

- Resolved #147997: user viewer: collaborators table.  

- Added both 'Lab Members' and 'Screen Associates' tables to UserViewer.

- Added show/hide feature to the above user tables and to the screen tables

- Added convenience methods to Lab and ScreeningRoomUser to traverse user associations.

- Resolved #147991: user viewer: screens table.  Added tables to UserViewer for both small molecule and RNAi screens, with "Browse..." buttons to see the screens in a full search results page.

- Added "Add Screen" buttons for specific screen types.

- Added ScreeningRoomUser.getAllAssociatedScreens() method, since its functionality was being duplicated in multiple places.

'screensaverUserRole' is no longer implied by other roles, since we do in fact want to allow this role to be exist independently of other roles, in order to enable/disable login privileges.

- Renamed screener-related ScreensaverUserRoles.

- Added root role, screensaverUser, which is required for logging in to
Screensaver (required for both admins and normal user).  This allows accounts
to be deactivated, and also provides a mechanism to support guest accounts, 
which can only view public data.

- Added explicit 'nonScreeningUser' role, since this role cannot be inferred
simply from the lack of the screener role

- Implicit 'screener' role is now hidden in UserViewer, per requirements

In UserViewer, default new lab affiliation category is 'HMS'

- ScreenDetailViewer now supports completely independent setting of labHead
and leadScreener; setting the labHead does not constrain the candidate
leadScreeners.  This supports creating screens for users that simply do not
have a lab/PI, or when the lab/PI has not yet been created in Screensaver
(and this requires special admin privileges).

- Screen now supports nullable labHead, which is required to support the above
independence of screen.{labHead,leadScreener}.

Added 'Add Lab Member' command to UserViewer 

In UserViewer, labAffiliation field is only visible for labHeads and then only in edit mode; this makes sense since the labAffiliation is always shown in the labName field.

- Resolved #147992 (user viewer: add new screen).  New screen can be added from the UserViewer, which auto sets the lead screener and lab head, and possibly the screen type, if it can be unambiguously inferred from the user's screening roles.

- Reverted back to non-null enforcement of Screen.labHead, as allowing a null labHead caused too many issues with new screen creation.  One effect of this is that a screen cannot be created for a user that does not have a lab head.

- Added some robustness, error-handling code to various viewer backing bean reload() methods.

- Resolved #147996: add new lab affiliation.  UserViewer now supports adding a
new lab affiliation

- UserViewer now supports LabHead creation, and only by users with
labHeadsAdmin role, via new main menu command, "Add Lab Head".  UserViewer 

- Renamed principalInvestigatorsAdmin role to labHeadsAdmin, for naming
consistency.

- LabAffiliation.affiliationName now mutable.  Default constructor public, to
allow new entity creation via UI.

- user.xhtml UI component now handles empty names

label change in ScreenDetailViewer

updated f+j init sql scripts to match schema update for rev 2536

Added schema migration script for rev 2536 (neglected to commit earlier)

ScreensaverUser.getFullName() is more robust, handling empty/null first and last names.

- Resolved #51828: add Lab entity

- Added new LabHead entity as a subclass of ScreeningRoomUser, to enforce type
safety of this special user type (also necessary for model unit test framework).  
'labHead' args in data model method signatures now take explicit LabHead type, 
instead of plain ScreeningRoomUser.

- A LabHead must now have a 'Principal Investigator' classification, while
non-lab head ScreeningRoomUsers cannot have the 'Principal Investigator'
classification.

- LabHead.labAffiliation may be null, while non-lab head
ScreeningRoomUser.labAffiliation must be null.  In other words, the
LabAffiliation is always stored with the labHead now, and a ScreeningRoomUser
can no longer have his own lab-independent affiliation.

- LabHead.labHead must be null, while non-lab head ScreeningRoomUser may now
have a null labHead, to support case where lab head has not yet been defined.

- UserViewer: lab affiliation is only editable when user is a lab head
(principal investigator); lab name only editable when user is a non-lab head

- Screen.labHead is now nullable, to allow screens to be created before the
required principal investigator user is created.  ScreenViewer now allows a
null lab head.

- Removed erroneous 'delete' cascade on Screen.labHead

fixed typo ("Havard", of all things to misspell!)

- Removed 'guestUser' role, replacing with 'screeningRoomUser'
- Added 'principalInvestigatorsAdmin' type, to be used in the future for preventing 'usersAdmin' role from adding users of type 'Principal Investigator'
- Renamed 'cherryPickAdmin' role to 'cherryPickRequestsAdmin', for consistency
- Renamed 'compoundScreeningRoomUser' to 'smallMoleculeScreeningRoomUser'
- ScreensaverUserRole now has explicit "displayable" name, allowing proper capitalization of "RNAi".
- Reinstated admin role validation in AdministratorUser
- Renamed schema table screensaver_user_role_type to screensaver_user_role
- AbstractBackingBean.isScreener() impl can now just check for the existence of the 'screeningRoomUser' role, since this role is implied by other user roles.
- Removed unused (and problematic) ValidatingScreensaverUserRoleSet class

misc AbstractEntityConverter enhancements

fixed staffBrowser.xhtml

Added 'User Classification' column to ScreenerSearchResults.

Resolved #148637: admin search result columns accessible by non-admins.
Made fix in TableColumnManager, which now controls access to these restricted
columns, rather than doing this at the UI/JSF layer.

- UserNameColumn now links to UserViewer, affecting all search results that
employ UserNameColumn

- Added many new columns to UserSearchResults, and made appropriate columns
administrative.

- Made appropriate properties in UserViewer admin-only.

- Moved staffBrowser.xhtml to admin/ subdirectory, to make admin-only.

- Added ss:labName facelet component
- Enabled links from ss:user and ss:labName to UserViewer

Added schema migration script

- Resolved #147990: basic user viewer properties.  

- Added User Viewer, providing viewing and editing of basic user properties

- Added 'Add Screener' command to main menu.

- ScreeningRoomUser default constructor is now public, to allow UI to create
an uninitialized entity.

- Removed ScreeningRoomUser.isNonScreeningUser property, as this fact can be
deduced (and controlled) via the user's roles.

- ScreensaverUserRoles:

  - now maintains role hierarchy, via 'impliedRole' property.
  ScreensaverUser.{add,delete}UserRole take advantage of this property to
  enforce that all implied roles are added and deleted as necessary.

  - added getDisplayableRoleName() method, for user interface support

  - added isAdministrative() method

- Removed ScreenDetailViewer.returnToViewAfterEdit data member, since no
longer serving any purpose

- UserSearchResults now link to UserViewer via new ID columns

- Updated version number to 1.6.3

developer config file updates

improved visual consistency of screen viewer publication input fields

Improved label for screen viewer attached file input

Added 'MIARE Document' AssayFileType, allowing MIARE information to be associated with a Screen in a quick & dirty way.

UI enhancements to ScreenResultViewer, which now makes use of ss:property and ss:collapsiblePanel facelet components for visual consistency.

layout changes to date fields in screenDetailViewer

Resolved #146417: increase sizes of screen viewer attached file input fields

Resolved #146223: screen viewer attached file problems

Resolved #146257: add 'primary screen report' attached file type

Added deprecation field to Well Viewer

- (Really) Resolved #145168: add 'deprecated' flag to well

- Added deprecated well columns to CPRViewer and WellSearchResults

- CPRViewer well deprecation warnings are now based upon lab cherry picks,
instead of screener cherry picks, since it is the lab cherry pick wells that
are actually being cherry picked.

- Made Well.isDeprecated a real property (no longer derived), since we want to
support searching/sorting on this property from the above data tables.

- Resolved #145168: add 'deprecated' flag to well.  

- Added Well.deprecationActivity, an AdministrativeActivity, to maintain details of the deprecation of multiple wells.  Well.isDeprecated is a derived field, based upon existence of deprecationActivity.

- AdministrativeActivity is now a concrete entity, with a 'type' property.  Allows us to create multiple types of administrative activities without having to create subclasses for each one (this does not scale well, due to table-per-subclass inheritance design)

- Added WellDeprecator command-line utility

fixed unit test

Resolved #145260: status item changes

Updated version to 1.6.2.

- All changes made to a screen in edit mode are now reflected in both the
ScreensBrowser and ScreenViewer when editing is complete.  Similarly, canceled
edits, which affect only the in-memory entity are *not* reflected in
ScreensBrowser or ScreenViewer.  All of this is accomplished by refetching all
data in the ScreensBrowser whenever the ScreenViewer has asked to view a
screen.  This, of course, fixes many bugs with not showing the latest version
of *persisted* data after saving or canceling screen edits.

- EntitySearchResults no longer responds to changes in the DataTableModel's
row index to set the current entity to be viewed, when in entity view mode.
This design was causing many problems, since the model's row index can be
changed by so many sources, and it's not possible to distinguish between
changes from user-generated scrolling events and programmatic changes.  In
particular JSFUnit updated the model's row index, causing problems exclusive
to test code.  Recent code changes (above) also caused performance problems,
since the multiple model row index changes were causing multiple screen viewer
and screens browser data reloads.  We now properly bind an actionEvent method
to the dataScroller component, allowing EntitySearchResults to update the
entity viewer *only* in response to user-generated scrolling events.

Updated newsViewer page for v1.6.1

Added required prefix text for small molecule screen titles

Fixed UI tests.

- Resolved #144055: screen search results refresh after screen edit.  
After adding a new screen immediately after login, the screen is now 
displayed in a properly initialized ScreensBrowser.

- ScreenViewer: Instead of using filtering to find and display a selected
entity in the EntitySearchResults, we find the entity in the already-loaded
search result.  This behavior fixed the above bug, but also provides more
intuitive behavior to a user that expects to be able to scroll through all
screens after finding a particular screen.
  
- ScreenDetailViewer notifies ScreenSearchResults to update itself after a
screen is edited or added.

- EntitySearchResults: renamed and simplified methods that scroll the data
table and methods that set entity to be viewed in entity view mode.

fixed 2481.sql

Resolved #143980: screen viewer: activities and cpr links should open in a search result.  ScreenDetailViewer now shows the (10) most recent lab activities and cherry pick requests in their respective tables and provides a "Show All" button that navigates to a full search results page for each data type.

Fixed bug with StatusDetailViewer.deleteStatusItem()

- Refactoring in screenDetailViewerBody.xhtml to use ss:collectionTable in
place of t:dataTable, for attachedFiles and statusItems

- Added code to support editable AssayReadoutTypes in ScreenDetailViewer, but
commented out for now, since Screen.assayReadoutTypes is a transient
property/collection (derived from ScreenResult)

Changed publication citation format to match format used in Cell (the journal).

Fixed bug in RNAiLibraryContentsLoader that occurred when reloading a library that had wells with existing silencing reagents.

ScreenDetailViewer: fixed bugs, made some cosmetic improvements, and performed some refactorings

- Resolved #143992: screen viewer: add billing info fields
- Some minor data model changes to BillingInformation, in particular, eliminating BillingInfoToBeRequested type

Added ui test for ScreenDetailViewer billing information creation

Corrected javadocs in ResultValueType

- Work towards #143992: screen viewer: add billing info fields.  Added a functioning Billing Information expandable section to ScreenDetailViewer, with a single amount field.

- Updated billing-related currency fields to use BigDecimal and a new 'Currency' property type in user interface

- property.xhtml now support Currency type

- Added CurrencyConverter

- Resolved #143981: new publication entity properties

- Resolved #50374: screen viewer takes pubmed id for new publication

- Added journal, volume, and pages properties to Publication.  Changed
pubmedId property to integer type.  Made all properties nullable.  

- Removed PublicationInfo class, replacing with Publication, which can now be
instantiated as a DTO.  Simplifies a number of publication data shuffling
operations.

- PublicationInfoProvider now also retrieves journal, volume, and pages.

Resolved #143993: screen viewer: publishable protocol

- Refactored ScreenDetailViewer and StudyDetailViewer, pulling up collaborator-related methods into StudyDetailViewer, thus recently duplicated code
- Removed obsolete, unused UserDAO.findCandidateCollaborators() method

- Updated ScreenDetailViewer user interface for adding and deleting collaborators 
- Resolved #128935: save screen: collaborators lazy init exception
- Resolved #64322: UI for edit collaborators is dangerous

- Resolved #143986: screen viewer: attached files content input.  Supports
uploading of attached files, as well as manual entry of file name and
contents.  Viewing of attached file contents handled via a simple file
download link.

- AttachedFile.fileContents is a now Lob Hibernate type, mapping to Blob sql type,
to accommodate true binary file attachments (such as .doc files, etc.)

- Added AttachedFile.fileType property, which is a controlled vocabulary 
defined by AttachedFileType

- Removed LetterOfSupport entity type and associated UI, as this is now
subsumed by AttachedFile w/fileType='Letter of Support'

- JSFUtils.handleUserDownloadRequest() attempts to guess and set mime type
when mimeType arg is null, providing more intuitive download/viewing behavior
from a browser

fixed unit test

- Resolved #143412: screen viewer always within search result context.  If
ScreensBrowser contains the screen that is to be viewed, the search result is
"reused" and simply scrolled to show the requested screen in entity view
mode.  This also allow any view that has a screen link, such as a Activity
Viewer and CPR Viewer, to return the user to the existing ScreensBrowser when
its Screen ID link is clicked by the user, preserving the screen search
result.  If the screen to be viewed is not in the current search result, the
ScreensBrowser is reloaded with all screens, and initialized to search on the
exact screen number (thus displaying  a 1 row search result).

- Added EntitySearchResults.gotoRowContainingRow(), to support viewing a
requested entity within the current search result context.  

- DataTable handles setting the current row index even before the view has
been constructed by JSF (i.e., before the DataTablUIComponement is bound).
This behavior was temporarily needed during implementation of above, but kept
in, since it may prove useful in the future.

- ScreenViewer now interacts with ScreensBrowser to display the requested
screen in the current screen search results, as described above.

fixed bug with Date{,Entity}Column that caused NPE when sorting on null dates

Menu screensForm defaults to 'findScreen' command, allowing Enter key to find screen, instead of creating a new screen

Cosmetic user interface enhancements to ScreenViewer and ScreenDetailViewer.

corrected schema migration script revision number

- ScreenDetailViewer now handles editing of StatusItems per requirements.

- Added new StatusValue 'Pending - NSRB' and 'Piloted'.  Renamed 'Pending' to
'Pending - ICCB'.

- Added a 'rank' property to StatusValue, to enforce ordering of associated
dates and to prevent mutually exclusive StatusItems from co-existing in the
same screen.  Screen.createStatusItem() enforces these business constraints.

- StatusItem now has only immutable properties.  Requires a Screen to always
delete and re-add a StatusItem in order to change it; this ensures business
constraints are validated by Screen.createStatusItem().

- Added unit and user interface tests.

- ScreenDetailViewer presents empty selection items in screen type, lab name, and lead screener fields when a new screen is being edited.
- Screen.{study,screen}Type properties are now mutable
- Reversed deprecation of JSFUtils.createUISelectItems() and added 'addEmptyItem' flag param

- Resolved #143411: add new screen command

- ScreenViewer edit mode now takes place in ScreenDetailViewer, as a separate
JSF view, simplifying the user interface during editing: no screen results,
collapsible panels, etc.  Activities and CPR tables are also hidden during
editing.

- Screen class updated to allow new, uninitialized entity object, needed by
user interface when creating a new screen:
    - allows for null leadScreener and labHead before Screen is persistent
    - studyType and screenType are now mutable, as long as they are null
    - Screen() constructor public
    - reinstated cascades to leadScreener and labHead; added tests to confirm
    that this works (not entirely sure why this cascades were removed in the
    first place, as they are quite appropriate)

- Added ScreenDAO.findNextScreenNumber(), used when creating a new screen.
Partial resolution of #75282 (provide a service to generate cpr_number,
screen_number, etc)

- ScreenDetailViewer maintains admin-view mode between viewing of different screens

- Added convenience methods to AbstractJSFUnitTest.  Added JSFUnit tests and
refactored existing *JSFUnitTest classes.

- Upgrade Spring Framework to version 2.5.4.

- Resolved #141439: library unloader out-of-memory.  
- RNAiLibraryContentsFileLoader now takes a plate range, allowing the library to be loaded incrementally, over multiple invocations.
- LibraryContentsParser.parseLibraryContents() now takes startPlate and endPlate params, although only RNAiLibraryContentsParser currently respects these arguments.

reverting changes from revision 2440, which did not resolve #142626 (screen viewer: dup performed by for new screening activities)

Resolved #142626: screen viewer: dup performed by for new screening activities

Resolved #142300: heatmap viewer plate scroll buttons invisible

Updated version to 1.6.1

Resolved #142581: activity viewer: empty end plate should imply single plate range

Resolved #142582: activity viewer: set focus to start plate after add

added support for 'list' property type in criterion.xhml

- Resolved #142369: add columns to screens browser to support reporting needs
- Fixed bug in Screens Browser Status and Status Date columns, which showed earliest status, not latest.  Updated Screen.statusItems to be a sorted set, and StatusItem natural ordering is now ordered primarily on date.  Reordered StatusItem enum values to produce more intuitive natural ordering of StatusItems with same date.

Resolved #142595: screen viewer: add cp screening error

- Resolved #142285: activity viewer fails cp plate activity with non-admin performedBy
- Added an explicit property access to activity.performedBy in viewActivity(Activity) method, as this is the only way we can get Hibernate to eager fetch the relationship; GEDAO eager fetch methods did not work, possibly due to an underlying Hibernate bug.  Other relationships are being eager fetched and cannot determine what is different about the performedBy relationship here, in particular because the defect only occurs for one particular activity instance.
- Added a unit test to verify/debug the issue, but the test does not encounter the same problem; leaving it in.

correct 2304.sql to work with our demo db on psql 7.4, which reflects the expected state of the database at this rev.

added logj4.properties file for demo site

Fixed CPRPlateMapFilesBuilder, which was erroneously outputting units for the volume value.  Now fixed to output uL value w/o units, and with 2 decimal places.

Updated news viewer content for 1.6.0 release

added schema migration script

cleaned imports

Resolved #99598: screen result download performance

Resolved #142146 (cpr viewer: view volume for unfullfiled wells when none unfulfilled).  HqlBuilder now properly handles 'in' operator with empty set.  Also disabled CPRV "View Unfulfilled Volumes" button when unfulfilled LCP count == 0.

- minor UI fixes for library importer pages

- minor UI fixes for library importer pages
- abbreviated 'Add Cherry Pick Request' to 'Add CPR' on menu panel

removed debug output

Library Contents Importer backing beans will log the library being imported to the user activity log

fixed a model unit test

- Refactored and fixed many model unit tests.
- AbstractEntityInstanceTest now chooses the min-arg create method when instantiating test entities, as this is more apt to create an entity with valid default property values (e.g., if business logic constraints take into account multiple properties).

- Resolved #141947 (add CPR fails)
- Added annotation to CPR.getEmptyWellsOnAssayPlate() to fix unit test, but unit test still fails

In UI, "Screen Summary" labels are now just "Screen"

Reinstated optimized code in SRDAOImpl.deleteScreenResult()

Fixed bug whereby ActivityViewer.save() returned to ScreenViewer, but ScreenViewer did not show new Activity in the Activities table, although Activity was persisted.

Fixed AbstractEntityInstanceTest.testGetterReturnsWhatSetterSet(), where it was *not* testing the value reloaded from the database, but rather just testing the in-memory value, which would always pass, of course.

Fixed a very subtle bug introduced by org.joda.time.contrib.hibernate.PersistentLocalDate.nullSafeSet(),which converted to the previous day for ranges [2008-03-09,2008-04-06] and [2008-10-26,2008-11-02].  Dates in these ranges would get decremented by 1 day when persisted to the database!  Note that these are the dates between the old and new U.S. Daylight Savings times, which changed in 2007.  Now using our own LocalDateType class instead of PersistentLocalDate.

- Improvements to user activity logging

    - logging aspect now logs current view name and the ID of the current
    entity, for entity viewers

    - Added EditableViewer interface, to add a getEntity() method to Viewer
    backing beans, for use by user activity logging aspect

    - Added self-proxies to StudyViewer and ScreenViewer to allow logging AOP
    to work on method self-invocations

- CPRViewer now implements EditableViewer interface, for consistency of
edit/save/cancel method names.

- reinstated "performedBy initially blank on new activity" requirement, which
was apparently lost in ActivityViewer.editNewActivity().

- Replaced more GEDAO.doInTransaction() calls with @Transactional annotation
in various viewer backing beans (for #69723)

- Resolved #120421 (add date to log output)
- Updated debug log4j settings

updated a ui test

Made ActivityViewer 'Plates Screened' table rows non-editable, to ensure plate numbers are valid for the library the row was originally added for.  Also corrected a bug that prevented a 'plates screened' row to be added if it shared the same start plate as another row, but had a different copy.

updated javadoc

Resolved #141439: library unloader out-of-memory

- LibrariesDAOImpl.deleteLibraryContents() now eager fetches required data, for performance. Added unit tests.
- Fixed bug in Well.setMolfile() that would prevent molfile data from being deleted
- LibraryViewer refactoring for #69723 (add transactional annotations and refactor service layer from backing bean methods)

Added ScreensaverUserTest, intended for previous commit

Resolved #120378: CPR.numberOfUnfulfilledLabCherryPicks is not correct.  Needed to decrement unfulfilled LCP count when a LCP is deleted.

- Replaced ComplexDAOTest with appropriately-named test classes.
- Added assert to test for #120378 (CPR.numberOfUnfulfilledLabCherryPicks is not correct) in CPRDAOTest

Renamed SimpleDAOTest to GenericEntityDAOTest

- Resolved #140791: activity viewer: activity date initially blank on create
- Resolved #141369: activity viewer: return to screen after editing new activity
- Resolved #140793: activity viewer: remove end plate validation.  Validation is performed at the point of adding a new "Plates Screened" row.
- property.xhtml supports 'required' attribute

- Resolved #140786: activity viewer: 'add new' buttons above table
- Resolved #140788: activity viewer: post-create behavior to add new activity
- Added a "Add Library Screening" and "Add Cherry Pick Request" buttons to the menu panel, making use of the existing Screen Quick Finder feature.

Resolved #140792 (activity viewer: auto-fill volume and replicate count on create)

Merged branches/release-1.5.3-maint 2356:2382 to trunk

- Resolved #130340 (data table sort order not maintained after filter)
- Fixed related and unrelated unit tests

Removed LibraryScreeningSynchronizer and RNAiCherryPickScreeningSynchronizer from ScreenDBSynchronizer, in preparation for transition from using ScreenDB Visits to using Screensaver Activities.

- Factored out duplicate operator and criterion UI code from searchResult.xhtml into operator.xhtml and criterion.xhtml Facelet components.
- Made some cosmetic improvements to the search fields

- Resolved #140668 (activities browser: clicking entity row that was last viewed has no effect)
- Fixed a search result scrolling problem with "Back to Search Results" command, which did not return to a page boundary
- Increased date type search field width

Resolved #140663 (date fields should accept 2-digit centuries)

- Moved logic that auto-copied assay protocol info from model layer to UI layer, as this is really an interface-level feature than a model business logic requirements. For example, it avoids unwanted behavior with ScreenDBSynchronizer.
- Moved existing test code into the UI test layer.

Updated version number to 1.6.0 and updated news

Merged branches/release-1.5.3-maint 2326:2356 to trunk

fixed unit test

- Resolved #140205 (activity viewer: assay protocol auto-set value)
- Screen.labActivities is a SortedSet and Activity.compareTo() now respects equals() contract (only considered equal if the same entity instance)

- Resolved #140350 (activity viewer: open in edit mode for new activity).
- Made LibraryScreening.platesUsed have a DELETE_ORPHAN cascade type, fixing bug whereby user-deleted platesUsed entities were not being deleted.
- In Activity Viewer, save and cancel commands navigate to current view, to avoid exiting the search results viewer when editing the entity from there

- Resolved #140349 (activity viewer: increase visibility of "For screener provided plates")
- Removed some unintended experimental code added in previous commit; fixed a UI bug in activityViewerBody.xhtml

- Resolved #140348: activity viewer: show library name for plates screened
- Resolved #140344: activity viewer: move comments to the bottom
- Resolved #140346: activity viewer: narrow performedBy list to users associated with screen

removed obsolete schema update script

correct schema update script revision number

- Resolved #69568: truncated dates in data model

- Entity model replaced Date type with LocalDate{,Time} types from Joda
library.  LocalDate Java type mapped to SQL Date type (was SQL Timestamp type)

- No longer need to use AbstractEntity.truncateDate() methods, as LocalDate
has the proper date/time resolution (calendar days only, not hours, etc.)

- Factored out dateCreated entity property in TimeStamptedAbstractEntity
class.  Made Activity, Screen, ScreenResult, ScreensaverUser inherit from
TimeStamptedAbstractEntity, thus adding dateCreated property to these
entities.

- Entity class constructors no longer take a dateCreated arg, as this is set
to the dateTime at the instant of instantiation.  Code can still call
setDateCreated() to override, if necessary.

- Removed business logic that set ScreenResult.dateCreated from date parsed in
screen result workbook file.  This value was redundant with the first
LibraryScreening date, and for that reason was not even being shown in the UI.
It also overloaded the meaning of the dateCreated property, which really
should be the date/time the entity was added to the database.  Removed
associated unit test and test data file.

- Removed ScreenResult.dateLastImported property, as dateCreated property is
sufficient.

- workbook2.Cell now returns Joda date types

- Updated CommandLineApplication to parse date args into Joda types

- No longer using Tomahawk inputDate JSF component, since it only returns a
java.util.Date type.  Better to use a simple text field, and optionally have a
date picker component populate that text field using client-side Javascript,
as this allows a JSF date converter to create a Joda date type.

- Removed hack from Criterion that truncated dates to calendar date
resolution.  No longer needed.

- DateUtil class removed.

- Fixed ScreenCreator Hibernate exception

corrected cmd line arg name in BoutrosAnnotationImporter

fixed javadoc

Resolved #140017: volume search fields not supported

- Well{,Copy}VolumeSearchResults now uses Volume column types.  Split max/min volume+copy columns into separate columns.
- Added Volume{,Entity}Column classes.

fixed 2343.sql for postgresql 7.4 compatibility

- Added Volume type, with VolumeType user-type mapping, and VolumeConverter JSF converter

- Updated data model and service layers to use Volume for all volume properties that were previously using BigDecimal.  

- All volumes are now represented in database in liter units.

- Updated property.xhtml to handle Volume type, allowing Activity Viewer and Cherry Pick Request Viewer to display and edit volumes with associated units.

- Updated ScreenDBSynchronizer's parsing of Visit volumes to use Volume type

- Updated unit tests to handle and use Volume type

- Removed Screening.estimatedFinalScreenConcentrationInMoles

- Changed to aqua-based color theme

- Image buttons now transparent, to reduce maintenance when updating color
themes

- Enhancements to dataTableNavPanel

  - Now a Facelets component, ss:dataTableNavigator, which accepts additional
  components

  - Removed "goto page", as I suspect no one uses this feature (and is less
  important now that search functionality exists)

  - Resolved #122181: location of navigation arrows on search results
  component changes; moved arrow buttons to left-most position and removed
  page number links

  - Resolved #139385: explicit "return to search results" button

- Moved Search Results "Export" button to dataTableNavigator

- ScreenViewer now supports adding new LibraryScreening and RNAiCherryPickRequestScreening activities.
- Added RNAiCPR.rnaiCherryPickScreenings relationship (was uni-directional)
- CherryPickRequests can no longer be deleted if they have associated rnaiCherryPickScreenings (enforced at both user interface at DAO levels)

- Activity.performedBy is now mutable
- ActivityViewer now supports editing of performedBy property

- ActivityViewer includes 'For screener-provided plates' property
- property.xhtml supports submitOnChanged attribute for Boolean type

Improvements to ScreenResultImporter memory usage:

  - ResultValueType.resultValues is now a 'bag' relationship type, instead of
  a 'map'.  Existing resultValues do not have to be loaded into memory in
  order to add new ones, allowing the ScreenResultImporter's 'append' feature
  to work effectively.  Also, Map type required a Map.Entry object for every
  element, which was wasteful.

  - ScreenResultImporter no longer eager fetches extant resultValues, as this
  no longer needed for the 'bag type relationship

ActivityViewer no longer sets default values for fields used to add a new 'Plates Screened' record.

- Changed ss:entityId UI component to ss:entityLink, which now takes an optional 'property' arg. 
- Fixed usage of entityLink to show screenNumber and cherryPickRequestNumber properties instead of entity IDs, which are not for user display in these cases

- Added Activity Viewer page, which currently supports all LabActivity types
(AdministrativeActivity types not yet supported).

- Lab Activity Browser data table now has ID column and links to Activity
Viewer

- Screen Viewer reworked to use new JSF UI components.  Layout reorganized.
Reinstated Activities table.

- Factored out UI code into reusable JSF components: ss:property,
ss:propertyGroup, ss:user, ss:collapsiblePanel, ss:collectionTable,
ss:entityId, ss:screenSummaryProperties, ss:editSaveCancel

- Screen.keywords is now a SortedSet, simplifying UI code.  User now enters
keywords in a single text field, as comma-separated list.

- All JSF converters are now Spring-defined (spring-context-ui.xml) and
instanitated as Spring beans.  This removes all converter definitions from
faces-config.xml.  Refactoring of collection-based converters, adding
CollectionConverter and SortedSetConverter, and updating ListConverter.

- Refactored edit/save/cancel backing bean methods into added EditableViewer
interface

- Updated data table nav panel color

Fixed date in newsViewer.xhtml

added lab affiliation category

Resolved #138455: CPR status fields viewable by admin only

added cherryPickAdmin role for Andrew Daab

minor io script change

minor io script change

CompoundPubchemCidListUpgrader improvements.  
  - Smiles to use is taken directly from Compound, rather than from associated Well(s)
  - Handles multiple error message XML structure

CompoundPubchemCidListUpgrader improvements.  Reduces requests for compounds in multiple wells.  Makes no changes to compound if an error occurs (for any its wells' smiles).

corrected log output

- Implemented PubchemSmilesOrInchiStandardizer.  Used by CompoundPubchemCidListUpgrader.  This should help find PubChem CIDs for the compounds that were previously not found (searching on smiles).
- Refactoring on util.eutils.Pubchem* classes to better handle differences in request/response between PUG structure search and standardizer services.

Resolved #131539 (tab order in well copy volume search result table)

added lab affiliation

- Temporary fix for #131182 (SM CPRV LazyInitEx when CPR.screen.screenResult != null)
- Simply commented out the UI warning that led to the LazyInitEx bug

Restricted access to 'Browse CPR' submenu items to readEverything admins, to be consistent with top-level 'Browse CPR' menu item

Resolved #130902: Browse CPRs by RNAi and SM

- Resolved #130341 (CPRV 'cancel plates' does not deallocate reagent volume), again (prev fix did not actually fix it)
- Re-instated the cached assay plate collections in CPR, for fear of performance degradation; also added comments explaining potential problems of using these methods

- Resolved #130341: CPRV 'cancel plates' does not deallocate reagent volume
- Removed cached collections of CPAPs in CPR, since cached collections
elements can become detached entities, while calling code expects the entities
returned by the respective transient methods to return managed entities
- Removed CPR from constructors that did not use it in CPLT and Screen
- Implemented CPRAllocatorTest.testCancelAndDeallocateAssayPlates() unit test;
also updated this class to test the numberOfUnfulfilledLCP property.

Removed CompoundCherryPickSynchronizer from ScreenDBSynchronizer, now that Screensaver has assumed responsibility for creating and managing SM CPRs.

Fixed a JSFUnit test.

- Fixed a JSFUnit test.
- Factored out AbstractJsfUnitTest from ScreensaverJsfUnitTest

- Fixed heat map viewer plate scrolling bug by fixing bug in UISelectOneBean.setValue()
- Fixed NPE in HeatMapViewer initialization, encountered by JSFUnit tests.

- Renamed ScreeningRoomActivity to LabActivity, updating all affected entity relationships, user interface references, etc.
- Performance improvement and simplifications to SchemaUtil.
  - Schema tables cached, improving full unit test run from 17 min to 13 min.
  - Removed 'initialize' feature that ran ICCB-specific database initialization scripts.  

cleaned imports

Removed unused methods from ReagentViewer

- Added Google Collections jar
- Simplified .project file

Moved some static methods from VPDEDM to EntityColumn

added f+j/sm-cpr-plate-copies.sh

change to exception message in LibraryCopyGenerator

moved resources/sql/initialize_database to f+j/initialize-iccb-database

reorganization of flotsam+jetsam/

Improvements to LibraryCopyGenerator, allowing plate numbers and ranges to be specified on the command-line.

fixed run.sh

In CPRV, added instructions on how to update the status of cherry pick plates.

- Resolved #124986: SM CPRs edge wells empty.
- Merged "requested empty wells" and "required empty {columns,rows}" concepts into just "requested empty wells".
- Both RNAi and SM CPRs default to having outer 2 edge rows and columns as empty

Changed plate type for compound cherry picks to ABgene (was Eppendorf)

- Updated javadocs, mostly in model.cherrypicks package
- Moved EntityNetworkPersister to test directory

Fixed unit test

- Resolved #123740: search result export not respecting filters.

- Made DataExporter and TableDataExporter interfaces more generic, by being
parameterized on any class, rather than an AbstractEntity

- TableDataExporter now must be parameterized with a DataTableModel
type. GenericEntityExporter has been simplified, now that its
TableDataExporter interface ensures that a DataTableModel is used as its
underlying data source.

- Updated WellsSdfDataExporter and factored out a (singular)
WellSdfDataExporter, since the WellViewer can no longer use the same exporter
as WellSearchResults provides (since that exporter is a TableDataExporter that
requires a DataTableModel)

- EntitySearchResults now explicitly takes data exporters of type
DataExporter<EntityDataFetcher<E,K>>

- Fixed a bug in InMemoryDataModel(), where only fetch() without also calling
filter() would not provide access to the underlying data.

- Updated tests for Well{s,}SdfDataExporter.  Updated GenericDataExporterTest
to test that an explicit (descending) sort order is respected.

removed unused src/org/apache/ directory tree

- Resolved #124158 (heat map viewer always disabled).
- Removed RVT.isNumericalnessDetermined transient property, as it was creating confusing code and really is not a model-level property.  
- Updated ScreenResultParser to determine an RVT's numericalness without using RVT.isNumericalnessDetermined

Misc, behavior-preserving mods, suggested by FindBugs

stub class for nat prods parser test

reworking of compound library contents parsers and EutilsUtils classes:

- reworked EutilsUtils classes to remove dependency on ParseErrorManager. every error reported by
  the EutilsUtils class is fatal anyway, so replaced calls to PEM.logError with throwing an EutilsException
  - adapted EutilsUtils class users to the new API
- added NCBIGeneInfoProvider as a Spring bean in spring-context-services.xml; injected into 
  rnaiLibraryContentsParser in the same file
- refactored some common code in SDRecordParser for retrieving pubchem cids
- improved DataRowParser error message for "well not found" to include the well key
- replaced barely useful test for SDFileCompoundLibraryContentsParser with some real tests. these
  tests cover everything (i can think of) seen in a valid sd file.
  - shrunk down the input file for these tests to make the test run faster
- plug in a MockNCBIGeneInfoProvider in the RNAiLibraryContentsParserTest, to speed up tests, and
  remove ugly entanglement with third-party "web service"
- re-created a clunky test in a static PubchemMolfileSearch.main method as a real unit test

removed bogus main method

renamed NCBIGeneInfoProvider to NCBIGeneInfoProviderImpl, in preparation for extracting interface

fixed error in javadoc due to copy-paste

a failed experiment with smiles strings that at best provides an example of using Ruby's Regexp and MatchData classes

CPRV plate type field displays full name of plate type

fixed the gene info unit test to match with the latest information from NCBI on the test gene.

its possible that this whole test will be taken apart, as i consider how to handle unit tests that hit third-party services, but i figured i might as well check in this obvious fix anyway.

updated newsViewer.xhtml

Updated version, news for 1.5.3

Fixed bug where after viewing a CPR (via quick finder), then opening 'Browse CPRs', JSF 'duplicate component ID' error occurred.  Instead of trying to figure out the real cause of this, simply replaced the lab/screener data tables with full-blown searchResult UI components, since I knew this design was working for screenViewer/screensBrowser.  So, user can now search/filter lab and screener cherry picks from within CPR Viewer.

cleaned import (fixes compile error on orchestra)

- Resolved #122673: small molecule cherry pick request viewer:
  - CherryPickRequestViewer now supports SmallMoleculeCherryPickRequests.
  - LCP and SCP data tables show appropriate columns.
  - Plate type is now editable and set to default value upon creation.
  - Requested/approved volumes set to default values upon creation (SM CPRs only).  
  - Updated CPRExporter.
  - Misc bug fixes: CPRV LCP table respects 'show failed' checkbox, which wasn't the case until the user clicked the checkbox; misc NPE fixes
- Added CPR.assayPlateType property and CPR.defaultAssayPlateType transient property (as abstract method)
- Added CPR.defaultMicroliterTransferVolume transient property (as abstract method)

resolve rt #123466: reagent finder page take db reagent_ids

- skip "Reagent Source ID" column header lines
- trim any vendorName + ":" prefix

removed bin/orchestra/ss-build, as it is replaced by ant install build target; updated comments in scripts; fixed screendb_synchronizer.sh

pluralized 'ChemBank ID' in newsViewer.

Updated version, news for 1.5.2

- Resolved #113961: search result data exporting.  All search result pages now
have a default Excel Workbook exporter that will export the filtered data with
user-selected columns.  Added GenericDataExporter and updated
EntitySearchResults to use it.

- Resolved #101104: download well search results performance is for crap.  The
new implementation is fast.  A 30K library exports in 30 seconds (with all of
the columns provided by the old, hard-coded WellDataExporter).  Furthermore,
now that the user can filter and select particular columns, the user does not
have to be subjected to the worst case export size, if they only want limited
data.  A 30K library w/only basic well columns takes 5-10 seconds.

- Removed Excel support for existing WellDataExporter and renamed to
WellSdfDataExporter.  WellSearchResults maintains it SDF export option, but
its Excel export option is nowhandled by the generic exporter.  This means
that we lose the separate worksheets for Small Molecule and RNAi wells, for
better or for worse.  However, WellSdfDataExporter now respects filtering of
search results.

- Added more well, compound, and gene columns to WellSearchResults, allowing
the user to recreate the same exported data set previously supported by the
WellDataExporter.

- DataExporter.export now takes a DataFetcher instead of a collection of data,
allowing easier and more performant implementation of data exporters used with
search results.  In particular, since the DataFetcher already knows about the
filtering and ordering options of the search result, the exporter now mirrors
the configuration of the search result.  Added K (key type) generic param to
DataExporter class.

- Added TableDataExporter subclass to support exporters that need to know
about a header row containing column names (a DataFetcher doesn't provide this
information).

- Workbook2Utils.writeCell() now handles collections (no longer uses default
Collection.toString()).

- Refactoring and improvements to HqlBuilder to simplify sibling methods that
handle entity properties versus value collections.

replaced a call to FileParseErrorManager.logError with a log.WARN message when encountering an SD record without a molfile. it would be nice if
this info showed up in the ui, but this is a non-fatal error, and should not prevent the library from loading. calling logError here prevents the
library from loading. 

added refseq version to summary in BoutrosAnnotationImporter

- Updated Katrina Rudnicki's Screensaver account to use eCommons ID for admin account, and screenDB credentials for her screener account, per her request.
- Added Andrew Daab admin account.

resolved #120834: in Well Viewer, message about compound structures for nat prods and NIBR1

script i used to generate a natural products excel spreadsheet from the NIRB1 SD files

fixed up CompoundNameValueTable so that the links for chembank ids actually go to the chembank site

tweaked some debug messages

Removed explicit exception handling for concurrency and database exceptions from all remaining @UIControllerMethod methods, as UICMEHA now handles these.

Reinstated commented-out UI definition that was causing a JSFUnit test to fail.  Will need to fix the test now, but better to have the test break than to disable UI functionality!

- Renamed ConcurrencyFailureExceptionHandlerAspect to
UIControllerMethodExceptionHandlerAspect, and it now handles both concurrency
and general database exceptions

- Added AbstractBackingBean.reload() method that is called by UICMEHA to
recover from exception properly (actually, this was added in previous commit)

- Removed exception handling code in ScreenViewer UIController methods that is
now handled by UICMEHA

- Added a JSFUnit test for UICMEHA, using ScreenViewer as test subject.

added a main method to load a natural products library file from the command line

- Created ui.cherrpickrequests package and moved appropriate classes there from ui.screens
- Minor refactoring in CherryPickRequestViewer

- added jars for jaxws-2.1.3
- added a build rule, chembank-wsimport, to run JAX-WS tool for generating service classes to be used by client code, based on the WSDL
- wrote ChembankIdUpgrader, to retrieve ChemBank IDs for the complete compound collection
- checked in the generated JAX-WS classes, to avoid Eclipse thinking the project is broken for users that have not run the chembank-wsimport
  build rule.
  - added a README file in the directory containing the actual generated classes explaining why these artifacts are checked in

unit test fixes

fixed unit test

Instrumented Screensaver with JSFUnit.  Web user interface tests can now be defined and executed against the web application running within the Tomcat container.  A couple of sample tests are in ScreensaverJsfUnitTest.java See https://wiki.med.harvard.edu/ICCBL/DevelopmentGuide#User_interface_Unit_Testing

Eliminated redundant definitions of main.xhtml

commented out bidirectional reln management for well->rv, for a different reason than it was commented out before. this is just killing screen
result importer performance + mem usage.

quick script to fix problem with trailing spaces in the new compound names from BioactivesSubmissionNameGrabber. this has been fixed in the BSNG as well (previous commit), but this script will fix the data from a bad BSNG run.

trim trailing whitespace from compound names - this trailing whitespace is bad because it is ugly data, but also because we are getting duplicate names, one version with the trailing space and one version without.

working version. depends on data cleanup in files 11_ICCBBioactives1.xml and ICCBBio1.sdf. also depends on reloading ICCBBio1 library with
latest version of the importer code.

modified to trim the contents of the SD file value fields. none of these values make sense with whitespace at the ends, and I have found that sometimes the Vendor attribute has a trailing whitespace, which leads to problems when trying to match reagent identifiers from other sources (eg, BioactivesSubmissionCompoundNameGrabber)

ugly fix for menu items not displaying in IE. the hover is still broken - no fuschia on hover on IE - but I am not about to add onmouseover and onmouseout javascripts to fix this

second half of replacing soft link to main.xhtml with actual copy

the soft link doesnt actually work in tomcat. first half of replacing with an actual copy

second half of a commit of a hacky fix for people having problems using bookmarked /main.jsf links. fix is to add a hyperlink from /main.xhtml to /main/main.xhtml, and allow /main.jsf in the web.xml file. other possibilities to do this more cleanly, but none that were simple - all of them required writing java code, which is rididculous. another possibility is to put a redirect in the META for the old page, asking people to update their bookmarks.

i accidentally checked in multiple files for this commit.

- bin/run.sh & bin/io/screendb_synchronzer.sh: fix CLASSES var to point at the build. improved the code that locates requires libraries - nows skips jars in unused/ directories

- web/web.xml: partial hacky fix for people having problems using bookmarked /main.jsf URLs (see next commit for the rest of the fix)

fixed to point at the correct log4j props file

removed db passwords from file

MedicinalCompoundsStudyCreator now:
  - uses the plate/well info to find the reagent when the provided reagent ID is invalid/unknown
  - ignores duplicate annotations for the same reagent when the annotation value is the same, and generate an error if they are different

- Updated eager fetching strategy in CPRV.viewCherryPickRequest(), to avoid potential cross-product fetch performance problems (performance should be improved, but no profiling was done to confirm this)
- By sheer luck, was able to get rid of the workaround jps added for an apparent Hibernate bug that prevented (simply tried reverting to original code after other eager fetch changes were made, which seems to avoid the Hibernate bug)
- Fixed 'Show Failed Lab Cherry Picks' command, which broke during 1.5.0 updates
- Updated javadocs in GenericEntityDAOImpl to address how to avoid cross-product fetch problems when using eager fetching methods

a script to clean up garbage CPLT rows created by buggy ScreenDBSynchronizer.

my first ever ruby script! fun!

reinstated jps fixes for roundedDiv panels, that got lost in facelets conversion

some refactoring of recent facelets changes

fixed login-error.xhtml

More fixes for facelets/roundedDiv issues on orchestra.  All dataTableNavigator panels now have a blue background, which was done only because it was easiest solution for now to get things functional; can change back at anytime

Fix that should address facelets problem in dev/prod environment.  Apparently, roundedDiv's color property requires a "property editor" for the color value, but it's not being found in the dev/prod environment, so switched to using a EL expression that provides an actual Color object, avoiding String->Color conversion.  The underlying cause of what's going on here is not understood, but this fix should hopefully get things working on orchestra.

manual schema update for svn r2228 (previous commit). i forgot to check it in along with the code changes.

added property harvardIdExpirationDate to ScreensaverUser, and ScreenDBSyncher. completes resolution of rt #116357.

improved error message for duplicate cherry picks in a compound cherry pick request. this should help track down errors reported by the synchronizer

fix for rt #113324

this basically resolves the problem in the ScreenDBSyncher, so no new garbage CPLTs will be created. i still need to go back and remove existing garbage CPLTs.

Added some missing indexes that are needed on result_value for performance.

Some UI files that were intended to be added in previous commit.

- Replace Tiles/JSP with Facelets view technology
- Version incremented to 1.5.1

some personally useful minor tweaks to my config files

fixed to insert correct revision number!

- removed various SAVE-UPDATE cascades
  - removed all bi-directional cascades that I came across
  - removed one or both cascades in situations where two different
    relationships cascaded to the same entity
  - in general, removed cascades that do not represent a containment
    relationship
- fixed up unit testing method for persisting a network of entities to do the
  right thing for all the various types of relationships that can be found in
  the network. extracted network persisting code into EntityNetworkPersister,
  and moved from test/ to src/, since it may well have some non-testing
  applications.
- added explicit calls to saveOrUpdate() and reattachEntity() for situations
  where these operations are no longer carried out by the cascade.
- modified CPLT constructors to no longer assume that the associated Screen
  is the same Screen as the associated CPR. this assumption fails in the
  model unit tests. it's also a data model violation, and should be checked,
  instead of assumed. added a comment about adding a check, but deferred on
  adding a check right now, since it would break the unit tests. this will
  have to wait for using AOP to check for data model violations, so that we
  can turn off those checks in the unit tests.
- made CPR.legacyCherryPickRequestNumber immutable
- made CPR.setRequestedEmptyWellsOnPlate private, Hibernate-only method.
  provided CPR.addRequestedEmpty() and CPR.addRequestedEmpties() as public
  methods in its place.
- added DELETE_ORPHAN to the cascade from Library to Well, just in case we
  ever try to delete a Library - otherwise we would have to manually
  delete all the Wells in the Library first.
- changed ResultValue.resultValueId from a SemanticID String to an
  @GeneratedValue Integer, due to problems with equals()/hashCode() methods
  that occured when the entityId of the parent RVT changed. (RVT.rVTId changes
  when the RVT is persisted. problems with RV identity occur within the same
  persist operation.)
- removed RV.equals(Object), since we really should be using
  AbstractEntity.equals(Object) here.
- uncommented code in RV constructor to maintain the bi-directional
  relationship between RV and Well. (if I remember right, the cause of the
  problem that led to this code getting commented out was the issue with
  the RV primary key discussed above.)
- split RVT.setNumeric(boolean) into two methods - RVT.setNumeric(boolean),
  which does *not* determine numericalness, and
  RVT.determineNumericalness(boolean), which does. this fixes a break in the
  model unit tests, where the test was first calling setNumeric, and then
  creating a new RV for the RVT, which ended up having an inappropriate
  numericalness.
- removed the extra-lazy on RVY.resultValues, at @'s suggestion - it's not
  being used for anything.
- added DELETE_ORPHAN cascade to Screen.{cprs,annotationTypes}
- reworked CherryPickAssayPlate.compareTo(CPAP) to use getter methods, instead
  of directly referencing the CPAP fields, to fix a problem with Hibernate
  proxies. A writeup of what was going on here is found at
  https://wiki.med.harvard.edu/ICCBL/ATrickyIssueWithHibernateProxies

cascades removed:

- AdministrativeActivity.approvedBy
- CPAP.cPLT
- CPR.screen
- CPR.labCPs
- LabCP.screenerCP
- Compound.wells
- Copy.library
- CopyAction.copyInfo
- CopyInfo.copy
- Gene.silencingReagents (retaining the cascade in the other direction -
  from the point of view of loading or traversing libraries, the Gene is
  contained by the SilencingReagent. we could removed the cascade in the
  other direction as well, since it appears nothing relies on it.)
- Reagent.{wells,annotationValues} (Reagents exist independently of both Wells
  and AnnotationValues)
- Reagent.studies
- SilencingReagent.wells
- Well.library
- AnotationType.study
- AnnotationValue.annotationType
- AnnotationValue.reagent
- RVT.typesDerivedFrom
- RVT.derivedTypes
- ScreenResult.wells
- BillingInfo.screen
- BillingItem.billingInfo
- EquipUsed.screeningRoomActivity
- PlatesUsed.libraryScreening
- RNAiCPScreening.rnaiCPR
- Screen.{leadScreener,labHead,collaborators}
- Screen.reagents
- ScreeningRoomActivity.screen
- AdminUser.activitesApproved
- ScreeningRoomUser.{labHead,labMembers,screensCollaborated}
- ScreensaverUser.activitiesPerformed

Partial resolution of #118298 (CPRV 'View Well Volume' is slow)

added a little trick to prevent dropping a fully loaded database, or truncating its tables. this has happened to me one too many times, and recreating a fully loaded database is a multi-hour affair.

minor fix in GenericEntityDAOImpl.relationshipSize(Entity,String,String,String) - use HQL Query.setInteger(entityId) instead of Query.setString(entityId.toString()) for integer primary keys, since the toString version becomes a problem somewhere between Postgres versions
8.1.11 and 8.3.0.

- Code changes necessary to resolve #51206 (medicinal chemistry comments on a compound).
- Added ScreenDAO{,Impl}
- Refactored deleteExistingStudy() in MedicinalCompoundsStudyCreator and IccbCompoundsStudyCreator to  ScreenDAO.deleteStudy()
- Added unique key to AnnotationValue on (reagent, annotationType)

OrchestraPropertyPlaceholderConfigurer now handles missing .pgpass file or matching entry in .pgpass file

Resolved #119247 (screen result 116 import error).  RVI now handles compound key delimiter character that may occur in the vendor identifier.

fix for IE[67] display problem of menu rounded corners

Resolved #118954: Internet Exploder display problem with roundedDiv rendering.  Still not perfect on IE{6,7}, due to rounded corner problems, but main menu is now visible again.  Good enough for now, since we recommend Firefox!

eliminated excessive MyFaces log warnings during rendering of search results table

Resolved #118013 (make search result queries read-only).  Not a win for performance, based upon temporal profiling, but worth keeping in.

Resolved #118691 (screen result parser derived data headers constraint)

- Resolved #75932 (data header names: underscores to spaces)
- Resolved #118758 (show screen title in screen result data table column header popup)
- Fixed security bugs with columns in special Administrative group
- Fixed display bug for screen/study names containing ':' character

Resolved #118757 (remove SRA list from Screen Viewer)

removed JSF hack for setting initial focus on an input component; we now use <s:focus> instead

removed code that originally resolved #50373 (scroll page to top if messages exist), but that we're no longer supporting.

more rounded corners work

removed now unused tomahawk-1.15.jar

- Upgraded MyFaces Tomahawk components to 1.1.6
- Added support MyFaces Sandbox components
- Removed hacked replacement classes for HtmlDataRenderer and HtmlInputDate.  Not getting involved with this level of hacking anymore.
- Partial resolution for #100720 (ui color scheme): rounded corners! (on menu panel, collapsible panels, and table navigation panel)

Resolved #118276 (view library contents exception)

- Resolved #118020 (ReagentSearchResults search on "reagent source id" error).  Reagent.reagentId returns a ReagentVendorIdentifier type, which cannot be used when generating SearchResult filtering queries; needed a simple String property.  So added an "alias" property, Reagent.reagentIdString, which maps to the same schema field as Reagent.reagentId.
- Added some basic units tests for ReagentSearchResults.

Resolved #118016 (entity viewers not hidden when search result panel collapsed).  Moved collapsible panels from WellSearchResults and ReagentSearchResults to ScreenResultViewer and StudyViewer, respectively.

Resolved #118012 (reinstate "All" rows per page in CPRV)

updated javadoc

fixed inserts to not write to no longer existent columns is_pubchem_upgrader_{succeeded,failed}, chembank_id

moved some classes to their proper package

removed exclude_table_data.sh, whose functionality is better performed via postgres pg_dump and pg_restore utilities

removed make-tarball.sh, which was non-functional and unused

updates to ant4's cfg files

updated README.txt

- Replaced README.html with README.txt, simply linking to Screensaver Wiki installation page
- Added svn:keywords 'HeadURL Id' to files missing it
- Changed a few top-level files' copyright year to 2008

- Version 1.5.0, merged from search-result-redesign branch (2188)

- Re-engineered SearchResults, DataTable design

  - Supports database-level filtering and sorting of data tables, for improved
  performance and reduced memory usage for large data sets.  Data binding
  mechanism for table columns, supports columns bound to related entities
  (e.g. well.gene.silencingReagent), even across one-to-many relationships
  (e.g. screenResult.resultValueTypes) via relationship restrictions.
  Supports user-configurable columns via column visibility (hide/show).

  - In particular, design allows for cross-screen comparison and joining
  annotations from multiple studies.

  - SearchResults now a subclass of DataTable.  New hierarchy is :
    DataTable
      EditableDataTable
        SearchResults
          AggregateSearchResults
          EntitySearchResults

  - Design separates generic-data SearchResults from entity-savvy
  SearchResults, which know how to fetch/sort/filter entity data.  This ensures
  reusability of high-level classes.

  - Introduced DataFetcher class hierarchy, for encapsulating data fetching
  strategies (eager and lazy) from data table models.  EntityDataFetcher does
  heavy lifting for retrieving entities for EntitySearchResults, and supports
  usage by both InMemoryDataModel and VirtualPagingDataModel.

  - New RelationshipPath and PropertyPath classes, along with EntityColumn
  classes hierarchy, providing a "property path" (column->entity property
  binding), allowing dynamic generation of Hibernate data fetching queries for
  data tables.  This will allow an arbitrary set of TableColumns to be
  selected by the user, and from any related entities, while ensuring that all
  necessary entity relationships are fetched to provide the required data.  A
  TableColumn property path is also used to generate restriction criteria that
  can be used to filter and sort a search result at the database layer,
  providing an efficient alternative to client-side, in-memory filtering for
  large data sets.

  - Removed all special-case DAO and UI code that supported the Study Viewer's
  Annotations panel data table and Screen Result Viewer's Data panel data table.
  AnnotationsDAO{,Impl} removed entirely.  Functionality is naturally
  supported by reuse of ReagentSearchResults and WellSearchResults,
  respectively.  Also, removed the special case filtering options from the
  Screen Result Data panel, as SearchResult's filtering feature replaces this
  functionality more generally.  Removed ScreenResultData{Model,Table} class
  hierarchies.

  - Removed DAO code that supported Well{,Copy}VolumeSearchResults, as new
  design supports it naturally, in particular AggregateSearchResults

  - All SearchResults subclasses reworked for new design.  CPR Viewer data
  tables updated to use new design.

  - DataTable has more refined event model to minimize database operations
  needed to support user-initiated actions (re-sort, column visibility, etc.)

  - Refactored ui.table package, adding column, column.entity, model
  subpackages.

  - renamed TableSortManager to TableColumnManager

- Model changes:

  - AbstractEntity.equals() now uses dual strategy: equality is in terms of ID
  equality if ID is assigned, or in terms of object identity if no ID is
  assigned.  Newly persisted entities should be NOT be used in subsequent
  Hibernate sessions, as their equals() method will fail (see Javadocs).

  - ResultValue key is no longer a generated ID, but rather a semantic ID,
  derived from RVT and Well.  Improves insert efficiency and also provides a
  single table column that can be used to recluster the table on.

  - Well.gene property now a real entity relationship (no longer a transient
  property), allowing data tables to access this relationship w/o special case
  logic

  - Well->ScreenResult relationship added (was unidir), though ended up being
  unnecessary

  - Reagent.annotationValues now a Map instead of a Set, keyed on
  AnnotationType

  - AnnotationType.values now a Map instead of a Set, keyed on Reagent

  - ScreenResult now ensures it child RVT names are unique.  Screen now
  ensures its child AnnotationType names are unique.

  - added many indexes for performance

  - added getActivityName() to Activity, to allow for a human-readable name
  for each class in the Activity hierarchy.

- Misc changes:

  - Renamed WellCopyVolume to WellCopy.

  - renamed DataTableRowsPerPageUISelectOneBean to RowsPerPageSelector

  - Resolved #100152 (CherryPickRequestViewer: "View Well Volumes" results are
  inaccurate)

  - Major improvements to HqlBuilder; new db.hibernate package

  - Added HqlUtils class

  - Changed Query interface, replacing getQuery() with execute(), and returning
  the result rather than a Hibernate Query object.

  - Reworked LibrariesDAOImpl.loadOrCreateWells() to explicitly handle various
  Hibernate entity states of the library (transient, managed, detached)

  - WebDataAccessPolicy restricts RVTs based upon parent ScreenResult 

- Extensive updates to unit tests

updated Jenn Nail's eCommons Identity

Originally created by an erroneous svn operation

Last sync point to search-result-redesign branch

all output of screendb_synchronizer.sh is tee'd to stdout and screendb_synchronizer.out, ensuring that errors are emailed to admin when run via LSF cluster

added @SuppressWarnings annotation

Removed copy.downloads target from build.xml.  For #113327 (find some way to get the "Data Downloads" compound library files outside of SVN).

fixed import_screen_result.sh script

update to import_screen_result.sh to support incremental import (start/end plate, append flag)

- Resolved #113119 (screen result importer terminates prematurely 449).
Chalked it up to excessive memory usage, by both JVM and postgres.

- ScreenResultImporter and ScreenResultParser now support incremental
importing of screen result files, by plate ranges.  This allows the importer
to handle a "manageable" size of data, without running out of memory, or
overloading postgres (which otherwise is being asked to complete the entire
import in a single transaction).

- Rewrote LibrariesDAOImpl.loadOrCreateLibraryWells().  I no longer understood
the comments we had in there, and wanted to have the code be written in a way
that made it very clear of its behavior w.r.t the Hibernate session; that is,
how the Library passed in is used when it's in either a transient, managed, or
persisted state.  The impl no longer relies upon catching of a Hibernate
exception to deal with a transient library object, avoids removed use of HQL.

- ResultValueType.resultValues map changed from Map<Well,ResultValue> to
Map<WellKey,ResultValue>.  This avoids having to have a full Well object
available to index into the map, which should optimize code that accesses this
map.

- WellKey is now an @Embeddable, so that it can be used as key type of
RVT.resultValues map.

- ResultValue.resultValueId now uses a 'seqhilo' ID generator, to optimize
bulk insert speed.  Resolves #113472 (ResultValue insert optimization).

- Updated unit tests.

some auxiliary files for ICBG report generation

added a couple more assay categories

simplified the mapping file name

added small clarifying comment

tried to limit assay category names to 30 chars

fixed testUseOfExistingEntities to synch with a change in the RNAi library contents loader - now,
we remove old silencing-reagent well contents before adding the new contents, instead of adding
both sets of contents to the well. i cant remember what motivated this change, but there was a good
reason for it...

various updates to get this test working again. also, made the test a little more resistant to
change by using sting constants instead of duplicated string literals for the contents of the
errors messages

Fixed NPE, which prevented CPR plate map files from being downloaded, when at least 1 LCP was not mapped

massaged js163 screensaver.props files.
updated Open Babel version comment in multiple .props files.

added Zac Cooper as admin user

- pushed up a wide variety of code from PubchemSmilesOrInchiSearch to parent class
  PubchemPugClient, including newly extracted method getResultsForSearchDocument(Document)
- created new PubchemPugClient subclass PubchemMolfileSearch

removed stray blank line

replaced single Compound.chembankId with set Compound.chembankIds

removed compound.is_pubchem_cid_list_upgrader_{s,f}

modified to look for Vendor field in SD file records. overrides Vendor value specified in Library.
initial assumption that all wells in a library have the same vendor is wrong (counterexample is
ICCBBio1)

- gather compounds for upgrading based on whether the set of pubchem cids is empty
- removed usage of entity props Compound.isPubhcmeCidListUpgrader{Successful,Failed}. these stopped
being meaningful, and at this point a failure is defined as the compound having an empty set of
pubchem cids

the usefulness of the success/failure flags was much more important when the upgrader was not
running on the lsf queue, and prone to need to be restarted.

in preparation for removing the Compound.pubhcmeUpgrader properties. i may regret this later,
but probably not, and they are really ugly fields to have in such a high-profile entity

one-off to address rt #80402 - retrieve compound names synomyms from bioactives data. currently
a work in progress

added new package upgraders to reduce clutter by moving out code that it not strictly library
contents loader code

enhancement to allow for null passwords

replaced obsolete the "place correct log4j properties file" rule with "place screensaver properties
file in dot eclipse dot classes." to bring back in line with recent changes to project
configuration. i have a feeling i will still be missing the correct log4j.props file here, but
i will fix that when it becomes a problem for me. (so i have a testcase to test the fix)

minor reformatting to make it easier to copy-paste the Perl code in the opening comment

disabled schema manager buttons that potentially kill the database, as they are very dangerous. afaik they are no longer being used by anyone.

added config info and deploy script for demo web site

fixes to help jsp

turn off hibernate SQL debugging

added administrator Jennifer Nale

modified to get structures based on the inchi instead of the smiles. i should really add some command line args to this program..

removed unused instance datum _vendorIdentifier

added useful synonym "Vendor_Identifier" for the column header "Vendor_ID"

bug fix: output sd files should have vendor identifier in Vendor_Identifier column for them to be
in the same format as the original input sd file. (otherwise they may not reload correctly.)

untested change to SDFile compound loader to address rt issue #110424: library contents loaders should check for existing reagents before creating new ones. will test on dev.

fixed fix of rt#110422 to avoid data model violation

untested change to nat products loader to address rt issue #110422: LibraryContentsUnloader should set well.reagent to null. (too easy to leave for another day.) will test on dev.

untested change to nat products loader to address rt issue #110424: library contents loaders should check for existing reagents before creating new ones. will test on dev.

For ICCB Compounds Study, changed "undesirable" annotation type to "unsuitable", for consistency with the language used to describe the study.

MedicinalCompoundsStudyCreator:
  - Temporarily changed to be a non-shareable study, until Study owners provide real data and approve.
  - Added output of errors.
  - Fixed bug where column headers row was being read in as data (though was not added as an annotation value)
  - Gracefully skips sheets that do not appear to contain data (no error generated)

Changes to {Iccb,Medicinal}CompoundsStudyCreator to change wording of titles, columns, headers, values, etc., per feedback.

Fixed bug that causes ReagentSearchResults to show the same well contents for every well.

Added MedicinalCompoundsStudyCreator, for #51206 ("medicinal chemistry comments on a compound").

fixed egregious textual error in the help system

fixed bug where molecular mass failed to account for the weight of the hydrogen atoms

fixed a cmd-line flag in make-tarball.sh

bin/run.sh actually runs now (how did that previous mess of a shell script stay in that state for so long?)

- Resolved #108859 (compounds study).
- Created a simple study, "ICCB Compounds" with annotation types "Undesirable" (true/false) and "Comment", and with a single reagent (NIBR1 K839-0057).
- Studies are now identified using screenNumber>=100000 (previous hack was studyType==StudyType.IN_SILICO).
- Fixed a small bug in SearchResults, where reloading search results page after viewing 2nd row in a 2-element search result caused summary view to start at row 2, permanently hiding 1st row.

changed PubchemSmilesSearch to PubchemSmilesOrInchiSearch. no functional changes, only name and
documentational changes to reflect the fact that the exact same code used here can function as
either a SMILES search or an InChI search.

Mods needed for prev rev to compile

Added command-line app to create "ICCB Compounds" study, for miscellaneous compound annotations.  For #108859 ("compounds study")

Fixed bug in CPRV where Download command failed if CPR.screen.screenResult==null

Fixed bugs that prevented well copy volumes from being edited.

enhanced debugging method to write a newline at the end of its output, for readability

upgraded to cdk version 1.0.1

added a quick comment reminding maintainers that the assayCategory being set gets put into a CHAR(30), so that long values get truncated. (indeed, many existing values are getting truncated.)

- Resolved #107344 (CPR empty wells).
- CPR now allows arbitrary set of screener-requested empty wells to be specified.
- CPRV provides short-hand for specifying and displaying full columns and rows of empty wells (e.g., "Col:10, Row:B").

fixed .classpath file

- Resolved #108491 (CherryPickRequest browser). Added 'Browse CPRs' command.
Admins can now view all CPRs at once, and determine completed status, etc.

- Resolved #108493 (CPRV: add status flag for unfulfilled lab cherry
picks).  Added 'completed' status, unfulfilledLabCherryPicksCount

- Updated CPR:

  - added numberUnfulfilledLabCherryPicks property, to enable efficient queries

  - added {pending,completed}CherryPickAssayPlates() transient properties

  - changed isPlated() semantics to indicate status of *all* CPAPs being plated

- Moved jsp files for cherry pick requests to their own directory (outside of
screens dir)

- Fixed CherryPickRequestViewer bugs:

    - Fixed "dup entity in session" error that occurred when recording
    subsequent plates as plated/failed/canceled.

    - CPLT datePerformed field is set to a current date, to prevent null value
    complaints

- Fixed BooleanColumn problem w/SelectItems.  Boolean column type restricted
operators to = and <>.

- Incremented version to 1.4.1

- Fix to DataRowParser to reuse existing Reagents when importing library contents
- Fixes to unit tests in RNAiLibraryContentsParserTest (it has failing tests, but these were failing before changes made today)

- LibraryCreator now reports parse errors and does not succeed if parse errors are encountered.
- Factored out io.ParseErrors and io.ParseError interfaces from parallel, but otherwise unrelated, ParseError and ParseErrorManager classes.  This allows LibraryCreator and library contents importers to manage parse errors from parsers of different library types in a standard manner.  This allows a single ParseLibraryContentsException to be introduced and that is now thrown from all implementations LibraryContentsParser.  Being able to throw this exception allows each implementation to rollback any failed attempt at creating/importing library contents (this was not the previously the case).

jsp fix

changed compound structure viewer problem workaround message to send users to PubChem instead of "OpenEye Depict".  The OpenEye Depict URL is no longer working and now that we have comprehensive PubChem CIDs, the user can go there instead.

Version 1.4.0.  Updated version number, news, main page.

fixed typo in UI		

Resolved #106497: delete screen result performance.  Uses HQL to delete result value in batches, issuing only one SQL statement per result value type.

- Work towards #106497: delete screen result performance
  - eager fetch wells before deleting ScreenResult, since Hibernate will load them one-by-one
  - Made ResultValue an immutable entity and made RVT.resultValues be a non-optimistic-lock relationship, in hopes of Hibernate using this hint to delete them w/o first doing a versioning check.  Neither does what we want, but leaving them in.  The immutable entity designation may help performance elsewhere.
- Updated unit test for deleting screen result.
- Fixed a bug in SchemaUtil command-line app's command-line arg definitions.

EntitySearchResult ensures that "table filter mode" is always disabled when in "entity view" mode, since search fields would entirely hidden from user in this mode.

- Reworked filtering UI in searchResults.jspf.  User can now choose to view filtering options in a standalone panel or in table columns.
- Significantly improved and reinforced the logic for converting user's wildcard expressions to regex expressions.  Production ready now.
- Fixed "vocabularly" spelling error in method names
- Promoted BooleanColumn type to have its own special-case UI, instead of being a VocabularyColumn type
- ColumnTypes now know the set of operators that are applicable to them
 

screen search results defaults to descending sort order on screen number

removed erroneous import

fixed library_copy_generator.sh script, by adding cmd-line arg pass through

- Added a very basic Screening Room Activities Browser
- Updated User Search Results to place first & last name in same column for easier filtering
- Removed unused dao arg from UserNameColumn
- Fixed WebDataAccessPolicy for CPLTs

- enabled filtering on ScreenSearchResults, which was erroneously disabled
- minor tweaks to date filter UI component

- Added basic filtering capabilities to SearchResults, for in-memory data models only.  TableColumns now manage a Criterion object.  All table filtering is disabled in the UI for now, as the code has not been fully tested for production deployment.
- Added "typed" TableColumn subclasses (Text, Integer, Vocabularly, etc.) and updated SearchResults subclasses to make use of these classes.
- TableColumn generic class now also parameterized on column data type (previously only parameterized on row data type).
- AbstractEntityConverter updated to handle null values.
- Added JSF Converter classes for each new TableColumn subclasses.

Added a very basic Users Browser, with menu options for browsing screeners and staff.

Resolved #104958 (pubchem CID links not working).  Insertion of non-breaking spaces into column names are causing column-name-matching logic to fail.  Moved non-breaking space reformatting to NameValueTable.getCellValue(), so that only JSF EL expressions obtain this HTML-formatted string.

added library_copy_generator.sh script

updates/fixes to bin scripts

fixed jsp warning

Removed WellKeyToResultValueMap, which broke in a recent rev when ResultValue was updated to first-class entity with a Well relationship.  Updated dependent code to use Well instead of WellKey when indexing into Well.resultValues map.

Resolved #104118 (download SD file for well(s) is broken).  well_molfile.ordinal value in db was 1 instead of 0.

- Fix to Well for unit tests
- Added custom unit test for Well.resultValues (a Map)

Additional optimization to CherryPickRequestExporter (in addition to optimizations made in previous revision)

- Resolved #102846 (output formatted result values in cherry pick result download file).
- Enabled CPRV "download" command for all users
- Fixed Well.resultValues Hibernate mapping

- Further optimized ScreenResultDataModel to only load those ResultValues for
the ScreenResult currently being viewed.  Previously, fetching of
Well.resultValues could result in poor performance if many different
ScreenResults have result values associated with the well.  Also moved a
query-per-result-value strategy, since this was actually faster than
batch-fetching across all ResultValueTypes (postgres doesn't seem to handle
the IN() SQL operator very well).

- Changed Well.resultValues mapping to be a map (keyed on RVT) instead of just
a set.  This allows optimizations to be performed when loading a well's
associated result values for only a single screen result (or some subset of
result value types).

- Corrected screen result data table's "well type" column to be "assay well
type"

- In ScreenResultSortQuery reinstated all temporarily removed functionality
(sorting, positives and plate filtering in terms of new HqlBuilder utility
class.

- Added ScreenResultDataQuery

- Reinstated "excluded" result values in ScreenResultDataTables.

- Fixes to ResultValueType.addResultValue() for (hopefully) improved logic for
determining whether RVT is numeric.

- Fixes to ScreenResultData{Model,Table} subclasses.  Also removed "all" rows
option, since this will usually generate queries with unacceptable performance
(and was difficult to handle correctly in code).

jar file re-rename for #103907

Resolve #103907: LibraryCopyGenerator fails on copy name "D" by forcing an ordering of jars with conflicting jakarta-cli code versions. This fixes the problem when running command-line utilities from our shell scripts in bin/io

upgraded to commons-cli-1.1 from 1.0 to resolve #103907 (LibraryCopyGenerator fails on copy name "D").

log4j.properties.util now appends to screensaver.log, to maintain ongoing record of a command-line utility's output.

reinstated build.xml fix from rev 2010, which was overwritten

update screendb_synchronizer.sh to work with updated build system

fix to OrchestraPropertyPlaceholderConfigurer

fixes to orch screensaver.properties.util.* files

- OrchestraPropertyPlaceholderConfigurer now uses an explicit property to find db connection settings on orchestra
- updated screensaver.properties.web.{dev,prod} for above
- OrchestraPropertyPlaceholderConfigurer now uses UNIX user name for database connection, if no SCREENSAVER_PGSQL_USER property is set; this allows sharing of screensaver.properties files on orchestra by multiple users, without having to edit the user name.
- split screensaver.properties.util into a dev and prod version to take advantage of above

build system fixes

fixed output location of build-number.txt in build.xml

build-number.txt is generated and copied for all builds, not just web deploy; this allows the "install" target to provide the build-number.txt file as well.

- Reorganized resources/*properties*/ into cfg/ directory tree, renaming files as appropriate.  This will support future command-line app configurations that any user may run on orchestra.  It also just keeps "ownership" of files clear among developers.
- For safety (and less confusion), build.xml *requires* -Dscreensaver.properties.file=... for build targets that need this file.
- Renamed log4j.properties.resources to log4j.properties.file, and made changes in cfg/*/screensaver.properties.* files.

adding back some bin scripts that should've been committed earlier 

Overhauled build.xml
  - New "install" target builds a usable installation of Screensaver command-line utilities, replete with a bin directory that contains useful utility scripts from screensaver/bin and the always-useful run.sh.  Can be installed outside of screensaver/ dir
  - "ddl" target builds to build/ddl, and uses the compiled classes and libs local to that dir (no longer uses .eclipse.classes, as that didn't work on orchestra)
  - removed many unused properties and targets

Removed build/ dir.

Reorganized bin/ directory

Reorganized bin/ directory

reworked 2002.sql to convert result_value_type_result_value to result_value by means of table and column renaming instead of table create-copy-drop strategy, which choked postgres.  Only runs on posgres 8.x+.

[Note: Features in ScreenResultViewer's data table have been disabled
temporarily in this commit, as some implementation tasks remain.]

- VirtualPagingDataModel now separates the strategy used to determine the sort
order of its data from the strategy used to retrieve the data that is needed
on-demand.  Previously, the data fetch strategy had to consider both sorting
and all-column retrieval at the same time, which was inefficient as a full
db sort had to be performed for each page viewed.  Cached data is reused
between sorts (although we're not currently taking advantage of this, since 
DataTable recreates the data model each time a sort event occurs).  This
should resolve #86568 ("srv data table query performance on orchestra"), but
must verify.  

- VirtualPagingDataModel uses a single top-level entity type to contain all of
the data for a row; previously a column name-to-value map was used.  The new
design allows for more flexible column compositions, since the data fetch
strategy only needs to provide an appropriately initialized entity (network)
for each row, and does not concern itself with returning specific, hard-coded
fields.  This is much more Hibernate-friendly, as the data fetch strategy can
operate fully at the entity model level, rather than returning a projection of
a tuple.

- ScreenResultDataModel (and subclasses) updated for this new sort/fetch data
access strategy and entity-per-row data representation.

- ReagentSearchResults now uses VirtualPagingDataModel, for efficient viewing
of large reagent sets.  This allows StudyViewer to no longer eager fetch
reagents, allowing it to open much, much faster.  Addresses log comment in rev
1990, stating that ReagentSearchResults needs to be optimized.

- ResultValue is now a first-class entity (no longer a "value" type), with
real associations to ResultValueType and Well.  This allows HQL queries to be
formed in a natural way, where ResultValues can be found for a particular set
of Wells and ResultValueTypes.  Added factory methods in RVT.

- added ResultValue<->Well relationship, allowing ScreenResultViewer data
table to retrive ResultValue data via the data model (see
ResultValueTypeColumn).

- ResultValue validates its assayWellType is valid for its Well's wellType.
Resolves #55975 (verify assay well type matches library well type).

- In ResultValueType, changed resultValues<String,ResultValue> to
resultValues<Well,ResultValue>.  Kept WellKeyToResultValueMapEntry for now,
since it's pervasively used, but can probably get rid of it now.

- added Study<->Reagent relationship, for efficient querying & sorting of
reagents in a particular study (see ReagentsSortQuery).

- ScreenResultsDAOImpl vastly simplified, having removed all special-case SQL
queries used to obtain ResultValues

- Added db.Query interface

- Added GenericEntityDAO.runQuery() convenience method.

- Fix in LibrariesDAOImpl.loadOrCreateWellsForLibrary() to handle detach
library

- Fixed ReagentVendorIdentifier compareTo() method, which sorted in reverse order

- Added units tests for ScreenResultDataModel and ReagentsDataModel

- Added many new entity-creation methods to MakeDummyEntities, for testing

- Added DebugUtils.elapsedTime()

fixed 2000.sql

retracted schema update for rev 1997 (never applied to dev or prod) and replaced with 2000, reflecting new mapping for Well.molfileList; previous Well.molfileSet mapping created index on the molfile value, which was too large for postgres indexes.

schema migration script for rev 1997.

Resolved #103139 ("reinstate Well.molfile") and #97116 (well.molfile isnt lazy loading), reverting back to earlier hack of placing the molfile in a single-element set, but now via mapped via JPA annotations.  Updated special-case unit test.

fixed JSF EL error

reinstated build-number.txt file generation in build.xml, which was commented out by accident

- Resolved #101441 ("resume screendb synchronizer"). 

- Fixed out-of-memory problem with CompoundCherryPickDBSynchronizer, by explicitly 
flushing and clearing Hibernate session after each CCPR is processed.  Each
CompoundCherryPickRequest is now deleted and imported as a paired operation,
and related libraries, screens, and users are reloaded each time.
ScreenDBSynchronizer now fully operates within a single txn.

- Mods to ScreenDBSynchronizer and data model annotations to fix things that
broke with the schema upgrade:

  - session is now explicitly flushed after requesting deletion of an entity's
  children (via emptying of collection), resolving insert/delete ordering
  problems w/Hibernate (caused, presumably, by FlushMode.COMMIT).

  - added DELETE_ORPHAN annotations to 1-to-many containment relationships,
  and removed from 1-to-1 relationship types (not applicable).

  - relaxed business rules on LabCherryPick and source well type, to allow
  ScreenDB data to be imported.

  - extant BillingInformation entities are reused, instead of recreated, to
  resolve insert/delete ordering problem w/Hibernate

- Resolved #101551 (optimize CPR delete).  Made a bit faster by eager fetching
all data to be deleted for a CCPR.  Not terribly fast, but better and good
enough for the synchronizer.

- ScreenerCherryPick equals() and hashCode() defined to catch dup entities at
set.add() time, rather than flush time (allows ScreenDBSynchronizer to handle
the problem via log error)

- In ScreenDBSynchronizer, a failed Pubmed query nows generates an error, but
not a fatal exception, to allow completion even when Pubmed is on the fritz.

- Removed now-unused CherryPickRequestDAO.deleteAllCompoundCherryPickRequests().

- Added clear() method to DAOs.

added TODO comment

Minor log output fix to BoutrosAnnotationImporter

Changed LibrariesDAOImpl.loadOrCreateLibraryWells() to use persistEntity() instead of saveOrUpdateEntity().  jps and ant explicitly worked on this together earlier, and thought we had committed this change, but I guess not.

updated javadoc for persistEntity()

[This revision is not ready for production, as Well.molfile is disabled,
ReagentSearchResults needs to be optimized, and more testing need to be
performed.  It is fully functional, however.]

- Partial resolution of #97774 ("Well, Reagent, Vendor, AnnotationValue entities & relns").
Added Reagent entity.  Currently, just a placeholder entity (only has an ID)
to allow for proper data modeling.  The code was getting way too cumbersome,
trying to use Well as an implicit container for annotation values; in
particular, ReagentSearchResults was in need of too many (inefficient) DAO
methods.  In the future we will have Reagent manage a well's compounds and
silencing reagents (per #102278).

- ReagentVendorIdentifier is used as the entity ID of Reagent, via
@Embeddable.

- Added AnnotationValue many-to-one relationship with Reagent.  This allows
entity model-based access to a Reagent's annotation values, which had to be
acquired via DAO methods previously.  Also allows simpler
ReagentSearchResults, which was parameterized on Well previously (as a hack),
but is now be parameterized on Reagent; no longer calls DAO methods to obtain
annotation values for reagents/wells.

- Added Well many-to-one relationship with Reagent.

- ReagentViewer is now Reagent-based, rather than Well-based (a hack,
previously).

- Library contents importers now create Reagents on-demand, as necessary.

- In StudyViewer, replaced AnnotationViewer with ReagentSearchResults.
Deleted AnnotationViewer and its supporting data table classes.

- Consolidated Library.createWell() methods into a single method (affecting
mostly just test code).

- Removed 'parentWellOfInterest' property in GeneViewer and CompoundViewer,
since it was only used in displaying a message to the user.

- Updated/fixed unit tests.

- Added ScreenCreator, which was intended for earlier rev.

corrected javadoc

updated javadocs in Library

correction to 1986.sql

- Added ScreenCreator command-line utility.
- Added unique constraint on Screen.screenNumber

- Added LibraryCreator command-line utility.
- Well now validates at instantiation time that it is within the range of the library plates it belongs to
- service.libraries.LibraryCreator validates it is not duplicating an existing library
- CommandLineApplication provides type-safe methods for parsing date and enum types

Added LibraryCreator service.

- Made all relationships lazy loading, to address monster queries Hibernate
was generating to load various entities, causing both performance and memory
problems.  The @ManyToMany, @ManyToOne, and @OneToOne annotations in our model
did not have fetch=FetchType.LAZY, but this is required if we want Hibernate
to generate proxies for these relationships (or lazy collections in the case
of @ManyToMany).  The relationships did already have Hibernate's @LazyToOne or
@LazyCollection annotations, and one would think that Hibernate would be smart
enough to say "hey, they want this to be lazy!".  Nope.  Hibernate assumes the
JPA-standard default of non-laziness, unless the *JPA* @{One,Many}To{One,Many}
annotation explicitly has fetch=FetchType.LAZY.

- Made ContainedEntity annotation 'inherited', allowing contained entity types
that are also subtypes (e.g., {RNAi,Compound}CPR) to be instantiated via
parent's factory method).

- AbstractEntityInstanceTest mods, to fix failing tests that presumably
started failing due to additional lazy relationships (above):

  - Changed how getFactoryMethod() finds the parent factory method.  Looks for
  "create" prefix and the expected return type.  Necessary since we cannot
  always infer the factory method name by the name of the contained entity
  type (RNAiCherryPickRequest's factory method is
  Screen.createCherryPickRequest(), not Screen.createRNAiCherryPickRequest())

  - Fixed testGetterReturnsWhatSetterSet(), which broke in cases where related
  to-one bean is now being lazy loaded

  - Added support for Character test value

  - getTestValueForType() forces creation of RNAi-type screens, to allow
  Screen-CPR relationship to be tested (sometimes Screen was 'Other' type, which
  doesn't permit associated CPRs)

  - Fixed AbstractEntityInstanceTest.getPersistedEntity(), which broke when
  calling getEntity() on a detached proxy instance.

  - other minor refactorings, add'l debug log output

added new lab affiliation

- In Search Results pages, replaced 'title' text of data header columns with heavy-weight JSF popups, allowing full data descriptions to be viewed (regardless of browser).  Especially needed for annotation columns, which have long descriptions.
- ReagentsBrowser has data panel open by default, and annotation types panel closed by default

Reagents Browser now has an Annotation Types collapsible panel (courtesy of the Study Viewer).  

Moved annotation-related UI classes to new ui.annotations package

Reagents Browser now has user-selectable annotation type columns. (Code worked perfectly on first web app test cycle! :)

- Resolved #100953 (data model violation rvt.positives_count).  Set isPositives=0 when RVT is a positive indicator.  Made positiveIndicator prop immutable, since changing this post-instantiation can lead to an incorrect positivesCount.
- Init ScreenResult.experimentalWellCount to 0 to fix NPE

Resolved #100750 (CPRV lab/screener tables' nav bar sort menus not synched).  Needed immediate="false" on commandSortHeader components.

update comments

inc version to 1.3.1, updated news, for #100703 (CPR user-requested empty rows)

- Resolved #100703 (CPR user-requested empty rows)
- Fixed unit test for CPRPlateMapperTest, and added this test class into our project-wide test hierarchy

minor refactoring in ScreenResultViewer

DataTableRowsPerPageUISelectOneBean now gets the "all" rows value via a template method, allowing instances to reliably obtain this value in cases where it dynamically changes and might be lazy-calculated.  Relieves the client code from having to know when to notify DTRPPUISOB of changes to this value.  This eliminates a lot of complexity in UI code that lazy-builds table data models.

added LAZY annotation for one-to-one relationships ScreenerCP => RNAiKnockdownConfirmation and
BillingInformation => Screen

wrapped up documentation for upgraded model

removed unused methods

some UI fixes for bugs recently introduced (more to come)

fixed two breaks caused by changing SR.experimentalWellCount from int to Integer

change _experimentalWellCount type from int to Integer, for consistency with the rest of the model

removed two unused instance fields

- ScreenResultViewer data table now uses dataTableNavigator.jspf.  So now all
data tables in Screensaver now use dataTableNavigator.jspf, and newsViewer.jsp
proudly notes this fact.

- Resolved #99596 ("screen result data table # rows")

- Within ScreenResultViewer code (i.e., multiple classes), removed all calls
to DataTableRowsPerPageUISelectOneBean.setAllRowsValue().  Was causing too
many bugs w.r.t the timing of calculating the "all rows" count for the various
ScreenResultDataTable subclasses.  As it turns out, if we don't set this at
all, the value of DataTableRowsPerPageUISelectOneBean.getSelection() is null
when the user selects the "all" option, and null happens to be the value that
tells our ScreenResultDAO methods to retrieve all rows!  I'm not proud of this
tenuous coding practice, but damn it, it works, and I'm keeping it!

fixed a schema_history.comment

removed unused imports

It appears Hibernate is not respecting any indexes defined via @org.hibernate.annotations.Index in @Embeddable entities, but added the annotation anyway for ResultValue.isPositive property, to address #100829 ("need index on result_value_type_results_value.is_positive").  So this is not a fix.

removed unique constraint on ssuser (first,last,date created)

fixed a test

fixed dummy entity creation (for tests), to work with ScreensaverUser unique key constraint on first,last,date

made proxyable, just in case

added a variety of multi-column uniqueness constraints. i did a reasonably exhaustive search here, but i obviously could have overlooked some. i leaned against adding constraints when i wasnt sure, since overly restrictive constraints is just gonna be a hassle later.

awaiting @s feedback on two constraints checked in here, but my assumption is they are good. if not they are easy enough to back out.

resolves rt #74027

- annotationViewer.jspf now uses dataTableNavigator.jspf, for consistency.  (Talk about plug-n-play s/w development!)
- updated newsViewer.jsp to note improved data table nav panels.

slight grammar correction that prevents javadoc tool from truncating the summary sentence for the class javadoc early

Resolved #97780 (UI data table code consolidation).  CPR Viewer now makes use of DataTable.

removed ignores for all the @hibernate tags that no longer exist from the javadoc build rule

updated usage instructions documentation from XDoclet annotations to Hibernate annotations

added unique constraints for screensaver_user (login_id) and (first,last,date_created)

(see rt #74027)

dont allow empty strings for ecommons id (in preparation for eventual unique constraint on this column)

added a news item for screensaver-1.3.0 - object model upgrade

- Partial resolution of #97780 ("UI data table code consolidation").
SearchResults now implemented in terms of DataTable, which simplifies the
class considerably.  Still need to make use of DataTable in CPRV.

- Refactored SearchResults, introducing subclasses to separate orthogonal and
mutually exclusive features:

  - EntitySearchResults.  Parameterized on an AbstractEntity type. Handles
  "single" view mode for viewing full entity, one by one, in browser.  Handles
  data exporters.  

  - SearchResultsWithRowDetail (currently used for
  Well{,Copy}VolumeSearchResults).  Parameterized on a SearchResults type for
  the row detail object

- Simplified design that allows user to navigate from an entity browser to
viewer (for a selected entity), and back to the browser.  "Rows per page" has
a new "Single" option (in addition to row counts) that, when selected,
replaces the search result table with the entity viewer.  Viewers no longer
know anything about the browser, or the fact that they are being used in
"search results" mode.  Browsers now embed the viewers. Removed nav panels
from viewers.  Removed "showNavigationBar" property from viewer backing beans.
SearchResults much simplified by this change.

- Nav panel in entity browsers has been refined and matched functionality of
nav controls in ScreenViewer and StudyViewer (goto page, page links, 10-page
ff/rew).  UI is now more page-oriented, rather than mixing row and page
numbers amongst various UI components.  The only row number provided is for
"total rows" display.

- Replaced searchResultsNavPanel.jspf with dataTableNavigator.jspf

- DataTable improvements:

  - DataTable class parameterized on row data type.  Column types are now
  parameterized on same generic type as DataTable, instead of the less-flexible
  Map<String,Object> column-name-to-value map.

  - Added gotoPageIndex()   

  - getRawDataSize() -> getRowCount()

  - Can register and notify observers when resort occurs.

  - DataTable now has buildRowsPerPageSelector() method, allowing finer control over
  instantiation of the rowsPerPageSelector bean by subclasses (e.g. to set the
  default selection and the "all rows" value).

- Removed some unnecessary UIComponent binding properties in backing beans
(NameValueTable, CherryPickRequestViewer)

fixed some UI popup text

removed some unused imports

improved an annotation description in BoutrosAnnotationImporter

- removed the setWellType(EXPERIMENTAL) on the source well, since this is pretty dangerous, since
it places an awful lot of undue confidence in the quality of ScreenDB data.
- instead, report an error if the well type is not experimental, and skip that cherry pick. added
a comment about libraries with loaded contents, and a referral to the class javadocs to
ScreenDBSynchronizer on the issue

- uncommented some code that i didnt want to check in commented-out
- added class-level javadoc warning about the perils of running against a non-recent screensaver
database

renamed GenericEntityDAO.persistEntity to saveOrUpdateEntity. added new GEDAO method persistEntity that calls Hibernate Session.persist instead of Session.saveOrUpdate. the former is JPA-compliant,
and will follow save-update cascades at the time of being called. the latter is non-JPA-compliant, and will follow save-update cascades at the time the session is flushed, or earlier.

modified LibrariesDAOImpl.loadOrCreateWellsForLibrary to use the persistEntity method on the library, so that the newly created wells are immediately available in the session.
        

added a @SuppressWarnings

added logger

report an error if there is a cherry pick in screendb for a plate for which there is no library

minor and inconsequential fix to error message

updated to work against latest schema

oops\! how did this one get lost? scary\!

removed external tool builder crap

ignored no-longer-checked-in .externalToolBuilders

deleted .externalToolBuilders from checkin, since it is dumping the Eclipse project name underneath. which doesn't make any sense, since this file is sitting right there inside the project? anyway, i am not going to assume that other Screensaver developers are going to be working on Screensaver with a specific Eclipse project name, or for that matter, working on only a single Eclipse project at a time

renamed manual schema update for the schema upgrade to an appropriate svn rev num

merged schema-upgrade-2007 branch back to the trunk. branches/schema-upgrade-2007 is now defunct.

snazzied up getInteger() method - if it is a STRING cell type, then ge the string, try to parse it
as an int (using the Integer constructor that takes a String). i also trim whitespace off the String ahead of time using String.trim() just in case there are leading or trailing whitespaces.

unflag compound as failed when successfully rerun

changed HQL for retrieving compounds to upgrade to get compounds that failed in the first run. i should probably create some command-line flags for this tool to specify which kinds of compounds to upgrade, instead of just hardcoding the changes..

Temporary fix for bug whereby only first annotation type was selected when Study Viewer is first opened.  Requires expanding the Annotation Types panel by default, which isn't too bad, but is not the original behavior.

fixed up all the test suites so the main program textui.Runners actually work

UI improvements to StudyViewer annotation tables:
  - show/hide annotations controls now display in a popup "dialog"
  - link to popup dialog is shown above both table panels if both panels are open, otherwise shown in the panel that is open
  - full text of annotation type description is shown in popup for annotation values table column headers

Resolved #95037 (well volumes should be viewable by readEverythingAdmin)

For #81743 ("roles incomplete for some users").  ScreenDBSynchronizer now sets users' roles based upon the screen types she is associated with (rnai or small molecule).  Untested, but rather straightforward code.

Fixed bug whereby wells with no vendor identifer (e.g. empty wells) causes AnnotationsDAO methods to fail.

nearly inconsequential, minor refactorings

- Resolved #97464 (help text for reagent viewer & browser)
- In StudyViewer, removed top-level 'Annotations' collapsible panel, making its child panels top-level panels.

BoutrosAnnotationImporter replaced "&" delimiter with ", " in "siRNAs" annotation type.

BoutrosAnnotationImporter now encloses refseq and gene ID groups with square brackets and uses commas to separate both bracketed groups and group members.  This replaces the "&" and "+" delimiters used in the original data.

BoutrosAnnotationImporter now makes use of existing users in the database (for lab head, lead screener, and rnaiglobal user), instead of deleting and recreating.  LoginID is used to find existing users.  This preserves any changes that are made in the database to the user accounts.

- BoutrosAnnotationImporter places "# Predicted Target Genes" annotation type before other "predicted" annotation types.
- Removed 'Entrez' from 'Entrez Gene IDs of Predicted Targets' for consistency with other columns, and to shorten its name. ('Entrez' is still found in its description).	

Added "Intended Entrez Gene ID" annotation type to Boutros study.  The new annotation type is derived by looking up the entrez gene ID in Screensaver.  In some cases the Screensaver gene ID is out-of-date, which becomes apparent when comparing the "intended" gene ID to the "predicted" gene ID and seeing they are different, even when the reagent is listed as having "Same annotation as Dharmacon".  In this case we would expect both intended and predicted gene IDs to match.  NCBI querying shows that Screensaver's gene IDs have simply been replaced with the "predicted" gene ID, which is the gene ID Boutros/Horn retrieved via the refseq entry.

added the latest lab affiliation categories

added log info about the number of compounds in need of an upgrade

Reordered columns in {Screen,Study}SearchResults to make Title 2nd column after ID.  Moved less important columns to the right.

removed non-functional command links in StudyViewer (invoking caused exception)

Resolved #58661 ("right justify numbers in search results")

Another heuristic performance tweak in AnnotationsDAO

- Fixed bug in ReagentsSearchResults, to handle case where reagent had no annotations
- Performance tweak in AnnotationsDAO: now making use of findRelatedAnnotationValuesInParts() method

minor help text fix

Boutros siGENOME annotation data is now shareable.

UI stylistic tweaks to {Reagent,Well}Finder same-page help.

ReagentFinder enhancements:
  - help text
  - "one-per-line" rule, to allow vendor IDs w/spaces to work
  - default to Dharmacon; a hack, but one that will be appreciated in the short term
  - only show ReagentViewer (versus browser) if the user entered exactly 1 identifier (and not just 1 correct identifier)
  - minor refactoring

{Reagent,Well}Viewer now uses non-breaking spaces for annotation names, for a substantial improvement in visual layout.

- Resolved #97458 (annotations in reagents browser).  ReagentSearchResults now shows all extant annotation types. Kind of bulky, but a short-term measure for release of Boutros data to RNAi Global consortium (may improve a bit before then, but ideal implementation requires #97780, #97460).
- Added Well.reagentVendorIdentifier (derived property), replacing ReagentVendorIdentifier(well) constructor.
- Factored out a mapPoolWellToDuplexWells() method in LibraryPoolToDuplexWellMapper, potentially for use in showing pool-to-duplex deconvolutions in Reagent/Well {Browser,Viewer}.

svn props

update log4j.properties.web.ant4
	

Renamed more "Vendor Identifier" occurrence to "Reagent Source ID"

- Renamed "Vendor Identifier" in Reagent/Well Viewer to "Reagent Source ID"
- Refactored " " hacks into a new HtmlUtils class.  Still a hack, but now it's a properly encapsulated hack!

Prevented AnnotationTypes of a restricted Study from being viewed in ReagentViewer.

minor jsp mods

missed commits

updated messages.properties

Fixed schema migration scripts to work with postgres 7.x.

Removed indentation on StudyViewer annotation sections, to make better use of space.  Left ScreenViewer as is.

Fixed bugs causes by recent schema change.  Proof the ant4 doesn't always test his code before commit!

- Added Screen properties: isShareable, isDownloadable, url. 
- Study entity is restricted if !isShareable.
- Annotations and screen result not downloadable if !isDownloadable.
- Study URL shown as external link in StudyViewer (not yet added to ScreenViewer)
- BoutrosAnnotationImporter now provides Boutros and Horn loginIDs and passwords.
- WebDataAccessPolicy grants view access to non-shareable Screens is user is the lab head, study lead, or collaborator.
- AnnotationValue.numericValue changed to double, to prevent loss of precision that would occur with fixed-precision BigDecimal (we only use the value for sorting in the database side, so a double is fine).
- ReagentVendorIdentifier.reagentIdentifier changed to vendorIdentifier for consistency with Well.vendorIdentifier.

Added explicit database index on well.vendor_identifier.

Added "Annotations" section to Reagent Viewer and Well Viewer.  These annotations tables really need to be more sophisticated and should show the parent study, the full annotation type description, etc.  But just wanted to get something in place quickly that will suffice for RNAi Global usage.  Will improve table in the future.

- Added Reagent Finder, Reagents Browser.
- Occurrences of reagent vendor IDs in the UI now link to the Reagent Viewer.
- AnnotationValuesTable now has single column for vendor and vendor ID, to be consistent with the rest of the UI.

- Added ReagentViewer.  WellViewer is now implemented in terms of ReagentViewer, and wellViewer.jsp "extends" reagentViewer.jsp.
- AnnotationValue entity now (implicitly) links to vendor reagents via a ReagentVendorIdentifier custom component type, in place of just a vendor identifier string (which was not a complete foreign key)
- Annotation viewer and DAOs updated to make use of ReagentVendorIdentifier.  AnnotationValueTable now links Reagent Source ID column to Reagent Viewer.

- AnnotationValue now displays the text version of numeric values (for numeric annotation types).  This preserves the decimal precision.  The numeric value is now only used for database sorting.
- Fixed AnnotationsDAO bug.

- Added popup hints to column headers in screen result data table and annotation value table, showing meta data description in popup text
- Disabled annotation viewer's "download" command (not yet implemented and not yet approved by Dr. Boutros)

Do secondary sort on annotation values returned by AnnotationsDAO.  TODO added to do similar in ScreenResultsDAO.

- Added admin-readable comments fields to StudyViewer
- Another "Reagent Source ID" mod

Renamed "Vendor ID" to "Reagent Source ID", which I'm not happy with as a name, but this is what Caroline has asked for.  (I think "Reagent Vendor ID" would be better.)

In StudyViewer's Annotation Data table, removed "All" option from "rows per page" selection list

- Welcome text is conditionally filtered for non-screener users, to show only applicable bullet points.
- Made a fix in BoutrosAnnotationImporter (rnaiglobal user was not being deleted on re-run)

- More changes to BoutrosAnnotationImporter
- Added sub-commands to 'Browse Libraries' command, to browse only RNAi or small molecule libraries (requested by Caroline)

- Updated Boutros study's annotation meta data with feedback from Caroline & Laura.
- Refactoring of BoutrosAnnotationImporter AnnotationType building and parse logic.

Corrected some inexplicable SVN anomaly, which caused ScreenViewer.java to not get committed (or something).  *Eclipse was happy; no compilation errors.)  Latest working version now committed.  Not sure how many previous revs didn't have the latest ScreenViewer.java file.  Oh well.

- Incremented version to 1.2.0, in light of the new Study entity type, StudyViewer, and StudySearchResults.  Although the Study entity type is not a real data model type yet (it's still just a Screen), the upcoming, proper implementation will be part of the 1.2.x series.  Update Latest News with this announcement.
- Performance tuned the AnnotationValueDataModel by making it into a virtual paging data model (on-demand fetching).
- Created VirtualPagingDataModel class, to support on-demand fetching of table data.
- Refactored ScreenResultDataModel hierarchy to use VirtualPagingDataModel as base class.  This simplifies these classes quite a bit, in particular, FullScreenResultDataModel, which originally went through some hoops to implement virtual paging as a subclass of ScreenResultDataModel.  Nice and clean now.

- Added new user role 'guest'.
- Added security constraints for study-related views, allowing guest users to view libraries and studies, and nothing else.
   - updates in web.xml
   - moved studies to their own jsp subdir for web.xml security definitions
   - updates in DataAccessPolicy hierarchy
- 'rnaiglobal' user is now a guest, and can only view libraries and studies
- 'rnaiglobal' password removed from source code; now specified as an arg to the importer

- Added StudyViewer
- Added StudiesBrowsernn
- Added 'Browse Studies' command to menu panel.
- Added Study class, just as a stub parent class of Screen (no effect on database schema).  This provides some type safety in the code as we transition to the Study->Screen->IccbScreen hierarchy.

- Refactored ScreenViewer and related jsp files, in preparation for implementing a study-only viewer.
- ScreenViewer is now a StudyViewer->ScreenViewer class hierarchy, which are top-level backing beans for the ScreenViewer, and manage only page-global properties and actions.  All of the backing bean props and actions for the screen detail panel have been refactored into a StudyDetailViewer->ScreenDetailViewer class hierarchy. 
- screenAndResultViewer.jsp, which used to be the top-level page definition for the ScreenViewer, is now just screenViewer.jsp, which includes jspf files for each of the major subsections.  All of the JSF code for the screen detail panel is now in screenDetailViewer.jspf.

Corrections for previous commit 1807.  Added a missing Java file, renamed the schema update file.

- Added AnnotationViewer backing bean, with related data table and data model classes, and added annotationViewer.jspf subview to ScreenViewer.  Can now view annotations associated with a screen/study.
- Added AnnotationsDAO
- AnnotationValue now has vendorIdentifier property in place of many-to-many association with Well
- Renamed Annotation to AnnotationType
- Factored out screen results UI from screenAndResultViewer.jsp into screenResultViewer.jspf, simplifying that former.
- TableSortManager handles 0 columns, and null sort column
- Updated BoutrosAnnotationImporter

More refactoring towards supporting code reuse for the "annotation" table in ScreenResultViewer.   Factored out screen result data table methods into a class hierarchy: DataTable -> ScreenResultDataTable -> {FullScreenResultDataTable, PositivesOnlyScreenResultDataTable, SinglePlateScreenResultDataTable}.

added a readEverythingAdmin account for Tao

- Fixed bug refered to in rev 1791, in ScreenResultViewer.  Recreates SRV's TableSortManager when data header selections change, instead of calling extant TableSortManager's setColumns() method, whose resultant behavior was more complex.
- Some UI enhancements to the ScreenResultViewer's data header selection panel.

- Factored out the "data headers" table from ScreenResultViewer into an abstract MetaDataTable class and a concrete ResultValueTypeTable subclass.  This is work towards supporting code reuse for the "annotation" table in ScreenResultViewer.  There are is a columns-changed-update bug, but debugging has proven too difficult due to complexity of multiple, cascading observer patterns, indicating that refactoring is necessary.  Committing this mostly working version before upheaval.
- Added MetaDataType interface, implemented by ResultValueType and Annotation.  This allows MetaDataTable to work with a common base type for its column data.
- Moved classes for heat maps to ui.screenresults.heatmaps.
- Moved DataTableRowsPerPageUISelectOneBean from ui.screenresults to ui.table package

awesome bugfix: download well search results wasnt preserving the well order as given by the well search results viewer page

(also fixed a typo in a comment in build.xml)

awesome enhancement: clicking on plate column header sorts by well key, instead of just plate!

granted Sean Johnston cherryPickAdmin role. (also on production database)

Simplification of ScreenResultViewer prior to refactoring (will be factoring out common functionality between screen result viewer tables and upcoming annotation viewer).  Now uses MyFaces dataScroller component, instead of custom paging commands and methods.  Eliminated binding of rowInput UI component.

svn props

Fixed #87694 (search results nav bar working strange on viewer pages)

Removed ui.controller package (had trouble doing this in previous svn commit)

- Removed LibraryController, moving all of its methods to their respective viewer backing beans.  Made similar changes as when removed ScreensController, including making all viewer beans have their bean dependencies satisifed via the constructor.
- Viewers that can show the search results navigation bar now "remember" whether they were opened in search results mode (w/nav bar) or in "single view" mode (w/o nav bar).  And back button works!  Made possible by use of MyFaces <saveState> component and 'showNavigationBar' property in the viewer backing beans.
- Fixed #60472 (losing search results nav bar in certain navigation paths)
- ScreenViewer now places Data Headers, Data, and Heat Map collapsible panels as indented, subpanels of the Screen Result panel.

added script for quickly loading screensaver db dumps by ignoring big data tables, like result_value_type_result_values.

Fixed a unit test.

corrected svn rev for most recent schema update script

- Work towards #81215 ("integrate Boutros' siGENOME sequence analysis data") and #73861 ("RNAi Global Screensaver Deployment")
- Added draft versions of Annotation and AnnotationValue entities to data model.  Will probably rename Annotation to AnnotationType, move to a new package, and make AnnotationValue link to Well via vendor_identifier, but commiting as is, in order to get on with an unrelated task.
- Screen's instance data members have been reorganized to reflect upcoming Study-Screen-IccbScreen hierarchy.
- Added StudyType controlled vocabulary ("in vitro", "in silico").
- ResultValue now formats numeric values with 0 precision properly (i.e. w/o a decimal portion).
- Added io.rnaiglobal.BoutrosAnnotationImporter class.

screen result exporter now handles null assay readout type

svn keywords

cleaned comments, imports

Moved EntityReplicator, from resources/ to src/  (woops!)

make-tarball.sh: more advanced, but broken

quick command to build a tarball of screensaver project for distribution purposes

For open sourcing:
  - added sample database dump; dump uses INSERTs instead of COPY for improved database compatibility
  - INSTALL.html; hacked together from wiki page save

ant takes credit for the class

fixed bug in EntityReplicator, allowing multiple entity loading to work.

- Some final touches for open source distribution. 
- Commiting .classpath, as I intend to include this eclipse prj file with the distro, for whatever help it provides to 3rd party developers.

- upgraded version to 1.1.0, to mark open sourcing of screensaver
- fixed a bug whereby view a single well showed nav panel from previous well search results; well search results now reset when viewing a single well, though nav panel is shown in well viewer even if only 1 well is in search results.
- removed AbstractUIController.logUserActivity() method (an aspect takes care of this now)
- fixed some typos, comments, imports

- Updated license, notice files, attempting to improve compliance for redistributed software. IANAL, but we're probably close to full compliance on redistribution terms.
- Added a licenses directory for redistribution licenses.

EntityReplicator:
   - now accepts relationships "groups", allowing it pass multiple relationships to a single call of GenericEntityDAO.needReadOnly() (et. al.), improving entity loading efficiency.
   - made most log output debug level
   - add some javadocs

CommandLineApplication now returns multi-valued args as a List of Strings (rather than Object).

Added a EntityReplicator command-line utility, allowing a Hibernate entity network to be copied from one database to another (e.g. a Screen and its child entities, or a Library and its wells and well contents).

Corrected header comment block to work with SchemaUtil

SchemaUtil handles comments that contain semicolons by ignoring lines that start with a comment (this is a bit of a hack)

DotPgpassFileParser now handles missing .pgpass file just a tad more gracefully; removed 'Orchestra' reference

CommandLineApplication now supports cmd line options that have multiple argument values

- Eliminated ScreensController, distributing its methods them to respective
viewer classes, and combining them with respective existing methods in viewer
classes.  Still have to do same for LibrariesController, but one at a time!

- Removed SearchResultsViewer class hierarchy, which consisted of mostly empty
classes.  As SearchResults subclasses were also backing beans there was really
no need for the *Viewer classes themselves.  Also, all SearchResults classes
are now Spring beans and therefore no longer constructed by our app code (this
strategy relied upon coding conventions to ensure that they were properly
initialized as AbstractBackingBeans; it also caused our code to constantly
pass SearchResult references back and forth amongst controllers and viewers).
We now can only have 1 search result at a time per entity type; this
constraint will be lifted once we support conversational-scope backing beans.

- Started removing logUserActivity() calls; now performed by
UserActivityLoggerAspect, which logs any @UIController method calls; added
@UIController annotations where needed.

- all Spring bean UI classes are now initialized via constructors rather than
setters; this is safer, as errors will be thrown at instantiation time if a
Spring bean's definition is missing constructor args (as opposed to silently
missing a property setter).

- Performed fairly comprehensive testing of the UI and it looks solid; haven't
tested all importers yet.  And there are a few minor UI bugs that need to be
resolved, but better to commit now for safety and deal with these in a future
commit.

Removed unused JSF Converter and Validator classes.

Menu.logout() is now logged as a user activity

- Eliminated MainController, moving all JSF action methods to Menu backing bean.
- Moved some JSF action result constants from ScreensController to ScreensaverConstants.

added missing initial whitespace on each line in header comments (Consistency(ness) is next to Godliness!)

recursively updated svn props with HeadURL and Id keywords, to fix file headers where keyword substitution was not happening

Preparation for upcoming UI code refactorings:
  - upgraded to Spring 2.0.6, as it now allows us to declare ordering of our custom aspects that also happen to have Spring transactional advice (i.e., @Transactional methods)
  - removed {Libraries,Screens}Controller interfaces, renamed {Libraries,Screens}ControllerImpl classes to names of respective, removed interfaces; this change will ease upcoming refactoring of migrating controller methods to backing beans; we will be eliminating the Controller classes altogether.
  - (re)implemented UserActivityLoggerAspect.  Couldn't originally get this to work with Spring 2.0.2 (possibly just because we didn't have CGLIB2 dynamic proxying working right at the time), but in any case it works now with 2.0.6, logging any call to a @UIController method; will be replacing explicit user activity log statements with this aspect (for now, we have duplicate user activity log output)
  - implemented ConcurrencyFailureHandlerAspect, which will allow us to remove db-related exception handling from @UIControllerMethods, which then allows us to make such methods @Transactional (we can't catch db exceptions *inside* a @Transactional method, since most db exceptions are thrown on flush, *after* a @Transactional method executes!)
  - Environment page now has "Throw a ConcurrentModificationException" command, for testing ConcurrencyFailureHandlerAspect in webapp environment

updated comments

- More work toward resolving #69723 (transactional annotations to replace dao.doInTransaction)
- Updated some javadocs re:usage of DAOTransaction{,RollbackException}

First (test) stab at resolving #69723 (transactional annotations to replace dao.doInTransaction), which will be a long-term, background project.

mods intended for previous revision

- Reverted back to original POI/HSSF library, removing our memory-improving hacked version, thus avoiding having to deal with any redistribution & licensing issues (we now use JExcelAPI when memory usage is an issue with workbook I/O).
- Moved workbook library memory usage testing code to f+j

- Added GPL license text to source files that were missing it, including xml and web files
- Added LICENSE-Apache.txt and NOTICE.txt, per the Apache License, required for redistributing derivative works (we've made some bug fixes to MyFaces code)

- Javadoc updates for top-level classes/interfaces in model package (these shouldn't interfere much with the future merge of the model update branch).
- Added package-info.java for model package

- Javadoc updates:
  - some new and rewritten javadocs, mostly at class levels
  - assortment of new package-info.java files and overview.html
  - lots of minor cleanup of @param, @return, {@link} etc. tags to clean up warnings
  - corrected spelling mistakes (thanks to Eclipse's new comments spell checker)
  - build.xml javadoc target handles our custom @motivation tag and ignores @hibernate xdoclet tags
  - "@author ant" to shared ant/jps @author tags
  - left javadocs in model classes unchanged, to avoid excessive merge difficulties with model update branch
- removed unused methods:
  - CherryPickRequestDAO.deleteCherryPickRequest(CPR, boolean)
  - ScreenViewer.getScreenTypeSelectItems()
  - JSFUtils.createUISelectItems(*)
- Fixed unit test testDeleteCompoundCherryPick

updated some documentation

removed defunct distro scripts: parse-screen-result.{bat.sh}

- Removed Sun's code sharing agreement (our code is now effectively unrelated to Sun's original example code)
- On login failure, detail failure message is now output to log

updated resources/README.txt

Fixed insipid bug whereby dates returned by JExcelAPI are for GMT time zone, but DateUtils.truncate() truncates based upon default time zone, causing truncation of "morning-hour" dates to be 1 day earlier.  See http://www.andykhan.com/jexcelapi/tutorial.html#dates, which blames Java, but I blame the JExcelAPI for not handling dates as being within the default time zone.

Removed AllCherryPicksImporter, which is no longer needed (it was for a one-time import process).

added some javadoc package-info.java files

- All code related to MyFaces and Hibernate library fixes/updates has been annotated with "screensaver"-keyed comments, for easy identification of changes made.
- Added some README files to explain why these library fixes/updates exist.

- All code related to MyFaces and Hibernate library fixes/updates has been annotated with "screensaver"-keyed comments, for easy identification of changes made.
- Added some README files to explain why these library fixes/updates exist.

some README documentation updates for open sourcing

some minor cleanup and README documentation updates for open sourcing

updated some properties files

updated ant4 eclipse prefs

fixed to not try to modify an unmodifiable set

added an all-essential blank line before the "constructor and instance methods" comment

Added file intended for rev 1683.

Minimum usable well volume in cherry pick plate wells is now 3.00 microliters, per Stewart Rudnicki's request.  (I *knew* this shouldn't have been a hardcoded value!)

Moved RNAiCherryPickCopyGenerator out of test code tree and into src.  Modified and renamed it to reflect its more general usage of creating library copies for any library type (now takes a plateType arg).  Added a little documentation for the next poor sap that has to figure out how to use it.

Added LibraryCopyGenerator service, with unit tests and an command line app to invoke it.

fixed AbstractEntity.truncateDate() to actually truncate the time portion of the date, rather than round it, which could result in a date value of the next day (if time was was > 12pm)

initial working version of the reverse screendb synchronizer

oops, my previous commit was a change to the schema-upgrade branch that should have been on the trunk:

- when the RNAi Cherry Pick Screening visit in ScreenDB does not have a crp_number_for_cp_screen,
  log an error, so that I will notice when I breeze through the logs, and can work with
  Katrina to fix the problem
- modified getRNAiCherryPickRequest from the crp_number_for_cp_screen, to first try the
  legacyCherryPickRequestNumber, and then try the cherryPickRequestId. i added in the bit about
  trying the cherryPickRequestId in preparation for the new situation, after the reverse screendb
  synchronizer has run, when some of the RNAi Cherry Pick Screening visits will have
  crp_number_for_cp_screen values that are from Cherry Pick Requests originally entered into
  Screensaver (see rt#89732)

- Resolved #88217 (CPRV: "View Well Volumes" only for unfulfilled LCPs).
- Rewrote & optimized LibrariesDAOImpl.findWellCopyVolumes(CherryPickRequest, ...), eliminating use of HQL, and instead just making use of the data model entities.

Addressed #50154 (screen result download option for "single data worksheet") by explaining that it's not going be fixed, unless user runs across the problem.

- Fixed and improved unit tests (and test data files) for ScreenResultParser.
- Fixed io.workbook2.Workbook constructor

optimized ScreenViewre's download command, by eager fetching result values.

- Resolved #89651 (ScreenResultParser OutOfMemory on LSF cluster).

- Created analog of io.workbook package in io.workbook2 package, using
the JExcelAPI library ("jxl") as the underlying technology.  This
library is considerably less memory-intensive than the POI/HSSF
library, allowing us to parse 50MB+ Excel files with only a few
hundred MBs of heap (still a crazy ratio, but...).  Classes in these two 
parallel packages are similar, but not worth sharing code between them, 
as the intention is to get rod of io.workbook (i.e. POI/HSSF) altogether.  

- Upgraded jxl.jar from 2.6.3 to 2.6.5.  Appears to fix a date parsing
bug.  

- ScreenResultParser now uses io.workbook2 classes.  

- Renamed 'metadata' occurrences in ScreenResultParser to 'detaHeader'
(was a legacy naming) 

- ParseErrorManager now only creates an annotated error workbook upon
request.  This improves time and memory performance.

- Updated ScreenResultParserTest and test excel files, as necessary

- Added JExcelApiMemoryTest and POIMemoryTest classes.  Will probably
remove, but wanted to save in VC repository.

added values for new Compound fields to 02_initialize_salts.sql

initial version of the News Viewer page. amazing how many mods needed to add a single page! the style for the page is quite undeveloped, and there is only a single news item so far. maybe @ can add something about being back from vacation!  ;-)

added "Download Well SD File" function to the Well Viewer page

- added notice about the structure image generator occasionally failing, with suggested workaround
- updated the help docs in a couple places, removing/updating content about bugs that have been
  fixed/are in the process of being fixed

added two new lab affiliations

set default to false for the pubchem cid list upgrader flags

fixed the RNAiCPR side of the RNAiCPRequest<=>RNAiCPScreening relationship from one-to-one to one-to-many

fixed user activity log message

removed some earlier hack-related code in SearchResults, and added a warning comment to getDataTable()

SearchResults row detail viewing feature is now less half-baked; let's call it three-quarter-baked.  Row detail stays open between requests, allowing sorting on row detail table.  Also, command links now work in row detail table.  No longer storing row detail object in servlet request map, which was a zero-baked idea, but instead storing with SearchResults baking bean.

- SearchResults no longer makes use of MyFaces DataTable's expandable row detail features, since it didn't play nice with sorting or edit mode
- Instead SearchResults now supports a simpler, somewhat half-baked one-row-at-a-time detail view.   The detail is only shown for the current HTTP request.  This is truly good enough for now, because users didn't even request this feature.  Will improve per user requests.

Screensaver software development process improvement: 
  - Added an SQL script that can quickly relay to the developer all schema migration updates made to the schema that it is run against.  Wow!  
    - Output is formatted in "expanded" mode (making this script psql-specific), for improved human readability.  Awesome!  
    - Revisions are sorted in ascending order, placing the most recent migration revisions at the bottom of the output.  Incredible!
    - Date field is truncated to exclude intra-day time precision.  Sweet Jesus!
    - Can be invoked on *any* schema containing a compliant screensaver_revision table. That's Hot!
    - Output can be piped to any UNIX command line tool that accepts stdin for further processing and analysis. Yeah, baby!
    - Tested on PostgreSQL 8.1.4 on i686-pc-linux-gnu, compiled by GCC i686-pc-linux-gnu-gcc (GCC) 3.4.4 (Gentoo 3.4.4-r1, ssp-3.4.4-1.0, pie-8.7.8). Righteous!
    

- SearchResults now supports a per-row detail view, which can be expanded or collapsed on a per-row basis via a link in a new expand/collapse column (conditionally shown, iff SearchResults subclass supports it).  Data for row detail is fed by another SearchResults backing bean.  JSF code for row detail is a pared-down version searchResults.jspf in searchResultsRowDetail.jspf (couldn't get recursive inclusion to work due to <t:aliasBean? limitations).
- New SearchResults row detail feature sounds great, huh? Well, I'm probably going to remove it, since it has bugs when data table is resorted or edited, and I'm tired of debugging half-baked MyFaces component features.  I'm only committing to save the work I've done, in case it proves useful again in the future.  Maybe should've committed to a branch, but whatever.
- Added WellVolumeAdjustmentSearchResults class, as backing bean for row detail in WellVolumeSearchResults.
- Many UI classes updated to accommodate new injection arguments (daos, controllers, etc.)
- Data model updated to make WellVolumeAdjustmnet relationships with LabCherryPick and WellVolumeCorrectionActivity bidirectional.  Only the Hibernate mapping changed, not the schema.
- Added GenericEntityDAOImpl.makeQueryIdList() static utility method
- WellVolume now maintains a list of WellVolumeAdjustments, instead of just the count.

removed unintended CSS style

fixed schema update to work on older versions of postgres (such as found on orchestra)

improved error handling

fixed naming convention to use underscores to separate words screendb and synchronizer throughout

added an explanatory comment

totally reworked the upgrader to function better in the very slow-running environment that
PubchemSmilesSearch gives us:

- takes advantage of new database fields for pubchem_cid_list_upgrader status to precisely and permanently (at least, semi-permanently) track the status of the upgrader per compound - if it has run, and if it was successful
- no longer processes compounds by library - instead, process all compounds that the upgrader hasnt run on yet
- commits every 100 compounds

help make orchestra-standalone env a production env by using the orchestra.prod log4j.props

added schema changes to record some information about the pubchem cid list upgrader execution

cleaned extraneous blank lines and comments

WellVolumeSearchResults:
  - fixed bug in calculating aggregate consumed volume
  - now shows the particular copy having the min and max remaining volumes for a given well

inconsequential CSS mods to WellVolumeSearchResults

ant4's help page text editing privileges are hereby revoked.

More fixes to help text, which apparently slipped through QA process.

Resolved #87296 (SearchResultViewer edits lost after sort)

fixed typos in help text

Fixed bug whereby sorting by clicking column header did not work in WellVolumeSearchResults.  Required having single <h:form> element around both tabbedPanel, and its child tabs' content.  (Can't have nested <h:form> elements in JSF/HTML.)  So searchResults.jspf no longer defines the <h:form> element, but requires it including parent file to do so.

- Fixed #87508 (disconnect between cell values and cell actions in latest search results code).  Sort column header required immediate="false".
- I had actually *already* made this *single* change to my local code to fix the earlier known problem where sorting by clicking column headers would lose values in editable fields (in WellVolumeSearchResults).  I came across this solution via the time-tested RTFM method (MyFaces taglib docs).  It turns out that when I said I couldn't reproduce #87508 locally, it was because of this one, single, tiny, little, existing change I had made to my local code, though I had declared that I was testing for #87508 against rev 1598.  Well, not exactly! Man, oh, man--precision really counts.
- Currently, WellVolumeSearchResults still has some sorting/editing-related bugs, but these will not be visible to screeners, so commiting this one change to fix #87508.

fixed up opening comment

- Combined WellCopyVolumeSearchResults and WellVolumeSearchResults into a single JSP page, displaying each in tabbed panel pane.
- Due to this change, CherryPickRequestViewer's "View Well Volumes" command now shows both search results for its lab cherry pick wells.
- Renamed UI-level files and variables from WellCopyVolume to WellVolume, since we had to choose a single name for the combined search results page.  Server-side code maintains WellCopyVolume term usage, since this is the level at which search results are generated.
- Updated help page for WellVolumeSearchResults.
- Column header sorting has stopped working on latest WellVolumeSearchResults page.  Will fix in future commit.  (Will be abandoning MyFaces' auto-sorting in search results, since it also had problems submitting edited field values between sorting requests.)

added code to synchronize RNAiCherryPickScreenings. These are very similar to LibraryScreenings,
and I managed to reuse a lot of code by extracting LibraryScreeningSynchronization code into an
abstract superclass, and made the RNAiCherryPickScreeningSynchronizer inherit from there. i also
extracted the code for determining volume transferred per well into its own class, tho it is a bit
of a mystery to me why i did that. oh well, its a quaint, short little class.

about the RNAiCherryPickScreenings - i need the ScreenDB 'RNAi Cherry Pick Screening' visits to
have a non-null CPR number (visits.cpr_number_for_cp_screen) in order to do this job properly. i
have discovered that 20/30 such visits do not have a CPR number. i asked Katrina about it and i
imagine she well be able to just fill some of them in, but it won't always be so easy because these
cherry pick requests are no longer being entered into ScreenDB. so i might have to write a little
script to back-populate these, which in turn might lead to further changes to the
RNAiCherryPickScreeningSynchronizer to work this all out.

cleaned imports (this once fixes a compilation error on orchestra!)

sql migration script for WellVolumeAdjustment.microliterVolume which is now "not null"

- Added WellVolumeSearchResults, which is an aggregate result of WellCopyVolumeSearchResults, grouping by copy.  WVSR is not editable, unlike WCVSR.
- WellVolumeSearchResults shows both max and min remaining volumes for a well (across all of its copies), allowing admins to determine whether a well (and all of its copies) is depleted, or is close to it.
- Well{Copy,}VolumeSearchResults have commands (in their header areas) to switch between one another.
- Still need to do some work to make these 2 search results be useful when navigating from CherryPickRequestViewer.
- Removed default sorting on remainingVolume columns, as I found this confusing as a user.

fixed to account for new PubchemSmilesSearch API: return null List when there is an error in the request

added some scaffolding code to handle the inevitable partial restarts on orchestra

- refactored and commented
- made improvements in error handling:
  - returns null instead of empty list on error
  - added handling for newly encountered status value "unknown"
  - should now appropriately report all error types, and not just ones ive
    already encountered

Back-pedaled on decision discussed in previous commit that stated, "Keeping UI-related WellVolume* classes as is, since they are likely to become dual-use, where, for example, WellVolumeSearchResults will be able to show either WellCopyVolumes or WellVolumes (an aggregatation of the WellCopyVolumes for a given Well)."  In fact, I'm going to create separately-named search results for WellCopyVolumeSearchResults and WellVolumeSearchResults (the copy-based aggregate of WellCopyVolumeSearchResults).  So all UI related code from WellVolumeSearchResults to WellCopyVolumeSearchResults.  God Bless Eclipse's refactoring features!

initial version of a script to run the CompoundPubchemCidListUpgrader on the LSF. its almost about time to write a run_screensaver_main_class.sh\!