Changes

#2194 (Nov 30, 2010 12:04:23 PM)

  1. Fixed bug 1812: there were issues with previous fix. It now works correctly. — gtc.cole@gmail.com / detail
  2. See BUG 3078 on Bugzilla. — gtc.cole@gmail.com / detail
  3. Fixed the primary build targets of the three products. The "dist" targets of mm and architect, as well as the "assembly" target of wabit now work well. — gtc.cole@gmail.com / detail
  4. Deleted the build.properties file under revision control in mm because this should be under svn:ignore >.> — gtc.cole@gmail.com / detail
  5. Fixed a few tests' database usage, and removed testEquals since Project no longer overrides equals(). — joe.cardwell@gmail.com / detail

#2193 (Nov 1, 2010 12:36:34 PM)

  1. Fixed failing tests that started to fail when the munge steps order was changed in the munge process. The tests have been refactored so the position is now handled in one place in the munge process. — ThomasObrien95 / detail

#2192 (Nov 1, 2010 11:25:34 AM)

  1. A few small fixes to modernize test cases and make them pass. — joe.cardwell@gmail.com / detail

#2191 (Nov 1, 2010 11:07:37 AM)

  1. Updated some tests, removing a few unneeded classes and fixing method implementations in others. Some tests will still fail, but they will all compile. — joe.cardwell@gmail.com / detail

#2190 (Nov 1, 2010 11:01:44 AM)

  1. Stubbed out the match statistics panel so that it displays without causing errors, even though there's nothing to show currently. — joe.cardwell@gmail.com / detail

#2189 (Nov 1, 2010 10:38:08 AM)

  1. Removed usage of the old repository so the program no longer makes an unnecessary database connection. As a result, sessions are no longer associated with a specific connection. Running engines from the command line now uses a file instead of a database. — joe.cardwell@gmail.com / detail
  2. Bug 1812. Doesn't allow Retain Characters field to be blank. — fridgebuzz / detail
  3. Fixes errors and bugs. Cleaned up some classes — gtc.cole@gmail.com / detail
  4. These changes are the new match pool implementation. It changes it and the match records into SPObjects so they can be persisted. — gtc.cole@gmail.com / detail
  5. Needed to persist reference match records after potential match records — gtc.cole@gmail.com / detail
  6. Some more getting rid of INTEGER(10) — gtc.cole@gmail.com / detail
  7. Changes to tests to adapt them to the new match pool after changing some match record functions and contructors — gtc.cole@gmail.com / detail
  8. Fixed another zombie reincarnation of INTEGER(10). The project editor was fetching tables with setting UserDefinedSQLTypes. — gtc.cole@gmail.com / detail
  9. Documented variable. — gtc.cole@gmail.com / detail
  10. Kept a repository connection from being used when the program is first run, so running two instances of the program does not cause total failure. — joe.cardwell@gmail.com / detail
  11. Used magic to ensure extra inputs and outputs are not added when old files are loaded through the single-project XML importer. — joe.cardwell@gmail.com / detail
  12. LIKELY FINALLY GOT RID OF INTEGER(10)
    +
    Fixed a bug to with selecting unique indexes. — gtc.cole@gmail.com / detail
  13. Allowed the exit check to account for database sessions so that enterprise clients end properly when all windows are closed. — joe.cardwell@gmail.com / detail
  14. The transformation selection popup window button in the engine panel now updates to the model so correctly its correct. — gtc.cole@gmail.com / detail

#2188 (Oct 22, 2010 5:59:14 PM)

  1. Added an unsaved changes confirmation when the user closes a non-server DQguru session. — joe.cardwell@gmail.com / detail

#2187 (Oct 22, 2010 4:26:13 PM)

  1. Changed the munge processes to have the result step at the start of the list instead of the end. The resason for placing the result step before the rest of the steps is because it is treated as final and the differ in the enterprise edition was re-adding the result step when it was being moved positions due to other steps being added and removed before it in the list.

    The better solution would be to go fix the differ but this approach was taken to meet time constraints. — ThomasObrien95 / detail

#2186 (Oct 22, 2010 3:56:14 PM)

  1. The transformation selection popup window button in the engine panel now updates to the model almost correctly. Still working on it though. — gtc.cole@gmail.com / detail

#2185 (Oct 22, 2010 2:56:14 PM)

  1. The transformation selection popup window button in the engine panel now updates to the model more correctly. — gtc.cole@gmail.com / detail

#2184 (Oct 22, 2010 12:47:13 PM)

  1. The transformation selection popup window button in the engine panel now updates to the model correctly. — gtc.cole@gmail.com / detail

#2183 (Oct 22, 2010 11:02:14 AM)

  1. Engine log file details now saved to prefs instead of being persisted. — gtc.cole@gmail.com / detail

#2182 (Oct 22, 2010 10:26:13 AM)

  1. Added an init() call when a de-dupe result step is added and magic is enabled, so that steps created in a client actually have an input. — joe.cardwell@gmail.com / detail

#2181 (Oct 22, 2010 9:42:14 AM)

  1. Fixed a bug with translate groups. When you make a translate group and leave it with the default name, then make another, the second one will have the same name, but the validator on the form won't catch it. Now it doesn't happen. — gtc.cole@gmail.com / detail

#2180 (Oct 21, 2010 5:39:14 PM)

  1. Removed Refresh since it is no longer needed. — joe.cardwell@gmail.com / detail

#2179 (Oct 21, 2010 12:21:14 PM)

  1. Added some default values to munge steps that required them, and cleaned up the Google Address step so the doCall function is clearer. — joe.cardwell@gmail.com / detail

#2178 (Oct 21, 2010 12:13:14 PM)

  1. Synched the UI with the model for the checkBoxes in the engine settings panel — gtc.cole@gmail.com / detail
  2. Changed an annotation. — gtc.cole@gmail.com / detail

#2177 (Oct 20, 2010 3:08:17 PM)

  1. To enable collaboration whenever a user edits a field of the PlFolder when the user changes fields or editors the changed properties will be written to the model and then passed to the server where other clients will then be able to update the properties from the server. The save button is still used to create a new folder and also gives users a nice feeling that their changes are saved when it is clicked.

    This will be extended to other editors shortly. — ThomasObrien95 / detail

#2176 (Oct 20, 2010 2:30:14 PM)

  1. Fixed a bug where loading projects of other types defaulted to de-dupe in the MatchMakerTreeModel because the property type wasn't persisted before the creation of the TreeModel. Also added refresh button to engine settings panel. — gtc.cole@gmail.com / detail

#2175 (Oct 20, 2010 12:52:13 PM)

  1. Updated the UI Update Manager to have a nicer API. — ThomasObrien95 / detail

#2174 (Oct 20, 2010 12:17:13 PM)

  1. Make an abstractUpdateManager so as to accomodate for different kinds of JComponents. — gtc.cole@gmail.com / detail

#2173 (Oct 19, 2010 6:30:13 PM)

  1. Refactored out the listeners for keeping the UI in sync with the model for the FolderEditor. This update manager will be useful in other data entry panels and will be used in other places shortly.

    The update manager is not complete as it still needs to update the model when the view component loses focus. — ThomasObrien95 / detail

#2172 (Oct 19, 2010 6:16:15 PM)

  1. Made over a mass of munges, making manifest a method that uses magic to mar the malicious multiplication of mungestepinputs and mungestepoutputs. — joe.cardwell@gmail.com / detail

#2171 (Oct 19, 2010 5:12:13 PM)

  1. Munge step outputs no longer persist their usage values, and munge step positions are changed and persisted properly, so they go where you tell them to go. — joe.cardwell@gmail.com / detail

#2170 (Oct 19, 2010 3:49:55 PM)

  1. Returning the project as the object under edit to fix an NPE when removing a munge process. — ThomasObrien95 / detail
  2. The folder editor now listens for changes to its model and displays the changes correctly. If an incoming change is going to conflict with a change made by the user the user will be notified in the UI. — ThomasObrien95 / detail

#2169 (Oct 19, 2010 3:42:13 PM)

  1. Munge steps now obtain their values before being built in the UI, and IO connections are no longer lost when loading. — joe.cardwell@gmail.com / detail

#2168 (Oct 19, 2010 1:16:13 PM)

  1. Added an event being fired when setting the result step so it is persisted to the server. — ThomasObrien95 / detail

#2167 (Oct 19, 2010 11:29:13 AM)

  1. The cachable table needs to load its catalog and schema when it gets constructed as it is used to load the cached table when it gets added to the tree model. — ThomasObrien95 / detail

#2166 (Oct 19, 2010 10:44:14 AM)

  1. Changed handling of munge step creation and loading so that XML imports do not generate mutant steps with too many inputs and outputs. — joe.cardwell@gmail.com / detail

#2165 (Oct 18, 2010 6:09:13 PM)

  1. The munge components now update based on its model as the model changes. The saving of a munge process is less useful but will be removed to allow for collaboration. — ThomasObrien95 / detail

#2164 (Oct 18, 2010 4:31:18 PM)

  1. Updated the annotated constructor to take in a tableName to let projects load from the server correctly. The table name is needed while objects are being loaded before other properties are set. — ThomasObrien95 / detail
  2. Removing the equality based on project name. This was previously used as project names were unique for Hibernate. This was also removed so projects could load correctly basing their new unique identifier on the UUID. — ThomasObrien95 / detail

#2163 (Oct 18, 2010 4:06:42 PM)

  1. Added in methods for creating sessions specific to an enterprise server. Now we can connect a client to a server. — ThomasObrien95 / detail

#2162 (Oct 18, 2010 4:00:13 PM)

  1. Removing the actual connecting to the data source being used as a repository. I am pretty sure the database/dbUser objects are no longer necessary. — ThomasObrien95 / detail

#2161 (Oct 18, 2010 12:27:15 PM)

  1. Added a test to ensure that object trees are being properly saved and loaded. — joe.cardwell@gmail.com / detail

#2160 (Oct 18, 2010 12:18:13 PM)

  1. Improved saving and loading so the process is more intuitive. Extensions are recognized in the file choosers, and Opening a project then Saving it does not open a Save As prompt. Factored out the actual saving and loading to a new class outside the swing package. — joe.cardwell@gmail.com / detail

#2159 (Oct 18, 2010 12:07:13 PM)

  1. You can now create AddressCorrectionProjects again :D instead of getting nulls in the SQL script. — gtc.cole@gmail.com / detail

#2158 (Oct 18, 2010 11:29:13 AM)

  1. Code cleanup. — gtc.cole@gmail.com / detail

#2157 (Oct 18, 2010 11:18:13 AM)

  1. Finished creating the new NetworkConflictResolvers. — gtc.cole@gmail.com / detail

#2156 (Oct 18, 2010 10:33:50 AM)

  1. More finalizing of the ClientSideSessions and NetworkConflictResolvers — gtc.cole@gmail.com / detail

#2155 (Oct 15, 2010 7:05:51 PM)

  1. Refactoring the NetworkConflictResolver — gtc.cole@gmail.com / detail

#2154 (Oct 15, 2010 6:15:42 PM)

  1. Moved some more stuff to the library to refactor clientSideSession stuff. CookieStore is evil. There are still compile errors in the MatchMakerClientSideSession but those will be resolved by refactoring the NetworkConflictResolver. — gtc.cole@gmail.com / detail

#2153 (Oct 15, 2010 2:54:16 PM)

  1. When the menus were moved so they could be edited from an enterprise class the initialization of the new project types were missed. — ThomasObrien95 / detail

#2152 (Oct 15, 2010 2:48:13 PM)

  1. Fixed an NPE when running the community edition version of MatchMaker. — ThomasObrien95 / detail
  2. Setting the column merge rules after the caches are set as the column merge rules also set the cache names. — ThomasObrien95 / detail

#2151 (Oct 15, 2010 2:01:13 PM)

  1. Added the ability to add an external listener to the context for the enterprise edition. Now an outside listener can be aware of other when sessions are created and destroyed.

    The menu bar builder has also been refactored so it can be edited by the enterprise version. — ThomasObrien95 / detail

#2150 (Oct 15, 2010 11:58:14 AM)

  1. Better generic placeholder. — gtc.cole@gmail.com / detail

#2149 (Oct 15, 2010 10:44:13 AM)

  1. The library tests and apt jars are re-created every build execution and do not need to be committed. — ThomasObrien95 / detail

#2148 (Oct 14, 2010 6:54:13 PM)

  1. Added setName() s to ensure that every object will have some sort of default name so the server doesn't yell at us for name having them. — gtc.cole@gmail.com / detail

#2147 (Oct 14, 2010 6:32:13 PM)

  1. Some fixes and changes. Mostly getting server resources working and cleaning up stuff. — gtc.cole@gmail.com / detail

#2145 (Oct 14, 2010 5:55:13 PM)

  1. Corrected the duplicate method to include the child list and added null checks for loading the SQLIndex. — ThomasObrien95 / detail

#2143 (Oct 14, 2010 5:30:14 PM)

  1. Passing a boolean to tell the context to not load the pl.ini on every creation of the context as not all tests need a valid pl.ini. This significantly reduces the memory and running time of the tests. — ThomasObrien95 / detail

#2142 (Oct 14, 2010 4:47:13 PM)

  1. Renamed the hibernate session and hibernate session context as they are now core sessions and contexts instead of being hibernate specific. — ThomasObrien95 / detail

#2141 (Oct 14, 2010 4:30:13 PM)

  1. Merged in the spobject-conversion branch.

    The merged branch includes all of the changes to the underlying MatchMakerObject model so it will work with the Enterprise persistence layer. As the object model changed and we planned to replace Hibernate with an Enterprise server the Hibernate code was replaced and the projects can now be saved and loaded to XML. Since Hibernate was replaced setters for final fields were removed and empty constructors were removed to correct the model.

    Use the hibernate-to-xml project in the MatchMaker SVN repository to export a hibernate repository to a set of xml files that are equivalent to exporting each project with the export feature. The projects can then be imported into a new MatchMaker session.

    The src.org.json package is now in the library and no longer needs to be included in MatchMaker directly. — ThomasObrien95 / detail

#2129 (Sep 15, 2010 4:55:04 PM)

  1. Fixed an error in building example tables in the help menu. Now it no longer shows null types in the generated ddl script. — gtc.cole@gmail.com / detail

#2128 (Sep 15, 2010 4:43:13 PM)

  1. new architect jar — gtc.cole@gmail.com / detail

#2126 (Sep 13, 2010 6:45:23 PM)

  1. Fixed some test cases in Matchmaker. — joe.cardwell@gmail.com / detail

#2103 (Aug 5, 2010 3:45:56 PM)

  1. Added a bundle id to the OS X application, so that it works correctly with spaces. — mc3ander@uwaterloo.ca / detail

#2087 (Jul 22, 2010 3:44:46 PM)

  1. Merged SPObjectUtils into SQLPowerUtils. We don't need that many utility classes! — terryyiu@gmail.com / detail

#2078 (Jul 20, 2010 6:42:22 PM)

  1. Removed the throws clause from SQLTable's getPrimaryKeyIndex() method since some old changes made it unnecessary for it to call populateColumns() and populateIndices. — terryyiu@gmail.com / detail

#2072 (Jul 16, 2010 11:34:58 AM)

  1. Fixed classpath to use Java SE 6. — terryyiu@gmail.com / detail
  2. Updated DQGuru to compile with Java 6. Currently, it is in a non-functioning state and needs to be addressed. — terryyiu@gmail.com / detail

#2069 (Jul 14, 2010 3:40:50 PM)

  1. Updated TestingConnection with JDBC4 API method implementions so that the project can be compiled with Java 6. — mo.jeff / detail

#2052 (Jul 12, 2010 10:18:18 AM)

  1. Fixed the build.xml file to not use the download template because the DQGuru page is now managed by hand and the template no longer exists. — terryyiu@gmail.com / detail

#2017 (Jun 14, 2010 4:57:12 PM)

  1. My previous commit to upgrade the MySQL connector JDBC driver from 5.1.5 to 5.1.12 has undesired behaviour. It causes users' MySQL driver in any SQL Power application that uses it to go "missing". The user would have to change their user preferences to relocate the MySQL driver after the upgrade.

    After a discussion with Jonathan, we agreed the quickest (but not the best) fix for this (and for any future driver upgrades) is to remove the version number from the filename. Then for any future driver upgrades, we would simply replace the old driver with the new one, keeping the same filename. However, to maintain compatibility with older applications which expect filenames with versions in them, we must keep the legacy drivers in place. Thus, we would have at most 2 versions of the driver: the legacy driver (with version in filename) and the most recent driver (without version in filename). — terryyiu@gmail.com / detail

#2008 (Jun 11, 2010 4:39:38 PM)

  1. Updated the MySQL connector JDBC driver from 5.1.5 to 5.1.12. — terryyiu@gmail.com / detail