Timeline
05/17/08:
- 15:50 Changeset [267] by
-
using string.substring(0..n) instead of string.split()[0] because it is more light-weight
- 15:48 Changeset [266] by
-
added missing GPL headers
- 15:47 Changeset [265] by
-
added missing GPL header
- 14:15 Ticket #153 (DistributedEngine needs too much RAM) created by
- There is a large increase in memory-footprint when executing an …
05/16/08:
- 11:40 Ticket #152 (Grid-server allows only 10 connections) closed by
- fixed: fixed with r264
- 11:30 Changeset [264] by
-
fixed #152, increased maximum allowed concurrent sessions to 20
- 11:17 Changeset [263] by
-
removed binding that is not needed anymore (#152)
- 11:15 Ticket #152 (Grid-server allows only 10 connections) created by
- 01:02 Ticket #151 (GetScope(String name) method of the Scope implementation in ...) closed by
- fixed
- 01:01 Changeset [262] by
-
Merged fix of ticket #151 (r260:261) from trunk into 3.0 release branch
- 00:53 Changeset [261] by
-
Fixed ticket #151
- corrected error in GetScope(string name) and GetScope(Guid guid)
05/15/08:
- 22:40 Ticket #151 (GetScope(String name) method of the Scope implementation in ...) created by
- The method GetScope(String name) does not get the subscope named name, …
- 12:18 Ticket #149 (DistributedEngine throws InvalidOperationException after running for a ...) closed by
- fixed: Worked on stability of plugin-infrastructure (uncaught exceptions in …
05/14/08:
- 20:14 Changeset [260] by
-
fixed #146 (serialization of function-trees in linear form)
- 19:24 Ticket #146 (Functiontrees should be serialized in compact linear form) closed by
- fixed: implemented with r259
- 19:23 Changeset [259] by
-
implemented #146
- 16:46 Changeset [258] by
-
moved sleep in the results-fetching-loop to the top. It's better to sleep a little before the first try. (ticket #149)
- 16:37 Changeset [257] by
-
don't keep all zipped engines just in case we have to restart one of them. Instead keep the engine objects (we have to keep them anyway) and serialize - zip only if a restart is needed (#149)
- 16:09 Changeset [256] by
-
- added statement to remove results from the job-manager
- added a critical section in case EndExecuteOperation is called while any of the TryGetResult-threads is still running.
- added statements to make sure that streams to (un)zip serialized engines are closed and resources released
(ticket #149)
- 16:06 Changeset [255] by
-
closing wait-handle just to be sure that resources are released (ticket #149)
- 16:05 Changeset [254] by
-
removed caching of results in engine-store. kept deletion mechanism for expired results to delete results that are never retrieved from the grid-server (ticket #149)
- 00:21 Changeset [253] by
-
improved handling of closed or faulted connections in job-manager for distributed engine (ticket #149)
- 00:13 Changeset [252] by
-
minor change
05/13/08:
- 23:56 Changeset [251] by
-
bug-fixes in grid-client (#149)
also cleaned up code in the grid-client
- 23:25 Ticket #140 (ServerForm shows incorrect numbers for "Running jobs" and "Waiting Results") closed by
- fixed: fixed with r250
- 23:24 Changeset [250] by
-
fixed #140
- 23:12 Changeset [249] by
-
fixed incorrect sleeping-timeouts (#149)
- 23:06 Changeset [248] by
-
improved stability of distributed engine (#149)
- 23:05 Changeset [247] by
-
deleted unused interface IClient (#149)
- 23:05 Changeset [246] by
-
minor bugfixes (related to #149)
- 23:04 Ticket #150 (Re-queue jobs when a client takes too long to return the result) closed by
- fixed: fixed with r245
- 23:03 Changeset [245] by
-
fixed #150 (re-queueing of jobs that were not executed in time)
- 22:37 Ticket #38 (Monitor running jobs on grid-clients from the grid-server) closed by
- invalid: This ticket has been replaced by #150.
- 22:36 Ticket #150 (Re-queue jobs when a client takes too long to return the result) created by
- If the grid-server doesn't receive a result for a job after a …
- 22:21 Ticket #137 (Stopping the GridClient while an engine is running throws an exception.) closed by
- fixed: fixed with r244
- 22:21 Changeset [244] by
- 20:27 Changeset [243] by
-
changed GridClient and GridServer to "service"-applications that are automatically restarted when they crash. (see r242) (related to ticket #149)
- 20:24 Changeset [242] by
-
added support for "service" applications that are restarted automatically when they crash with an exception. (related to #149)
- 18:58 Changeset [241] by
-
Stability improvement.
also makes debugging easier (related to #149). Should be merged into the trunk after testing.
- added exception handling in PluginManager to graciously unload AppDomains when an uncaught exception is occurred
- added exception handling in the starter-form to display an error message-box
- 18:36 Changeset [240] by
-
removed start/stop buttons, server is started automatically right away (necessary for automatic restarts)
- 15:30 Ticket #149 (DistributedEngine throws InvalidOperationException after running for a ...) created by
- This happens at least for the Meta-GP engines of pspitzli. The cause …
- 15:23 Ticket #30 (Make clients reconnect to the server periodically in case a connection ...) closed by
- worksforme: This issue was fixed together with another issue in an earlier changeset
- 15:11 Ticket #148 (Function-evaluator should copy code and data only once per evaluated model.) closed by
- fixed: fixed with r239
- 15:10 Changeset [239] by
-
fixed #148
- 15:01 Changeset [238] by
-
used Debug.Assert to enable tree-validity checks only in debug builds
- 14:44 Ticket #148 (Function-evaluator should copy code and data only once per evaluated model.) created by
- Right now the function-evaluator copies the lists on each evaluated …
- 14:37 Ticket #147 (Scaling factors should be saved together with the dataset) closed by
- fixed: implemented with r237 by moving scaling code and data from …
- 14:36 Changeset [237] by
-
implemented #147
- 13:40 Ticket #147 (Scaling factors should be saved together with the dataset) created by
05/12/08:
- 22:31 Changeset [236] by
-
minor changes
- 15:33 Changeset [235] by
-
fixed bug in function evaluation.
if a tree had the same size as the tree that was evaluated just before the evaluator evaluated the last tree again.
- 10:38 Ticket #143 (Exporter from HL dataset to libsvm data-format) closed by
- fixed: Implemented with r234
- 10:37 Changeset [234] by
-
implemented #143
- 08:00 Changeset [233] by
-
removed export button and attached the contextMenuStrip directly to the dataGridView instead of each column. (preparation for #143)
05/11/08:
- 12:33 Ticket #144 (Automatically scale variables in a dataset) closed by
- fixed: implemented with r232
- 12:33 Changeset [232] by
-
implemented #144
- 10:40 Ticket #145 (ProgrammableFunction not working with the new evaluation code) closed by
- fixed: fixed with r231
- 10:39 Changeset [231] by
-
fixed #145
- 09:00 Ticket #139 (Implement fast evaluator for GP function-trees) closed by
- fixed: * cleaned up files and fixed compilation problems with r225 and r226 …
- 08:58 Ticket #146 (Functiontrees should be serialized in compact linear form) created by
- 08:58 Ticket #145 (ProgrammableFunction not working with the new evaluation code) created by
- 08:56 Changeset [230] by
-
deleted old branch for ticket #139
- 08:55 Changeset [229] by
-
merged changes (r201 r203 r206 r208 r220 r223 r224 r225 r226 r227) from branch ExperimentalFunctionsBaking into the trunk. (ticket #139)
05/09/08:
- 16:22 Changeset [228] by
-
bug fixing in DistributedEngine and Grid-Infrastructure
- 15:23 Ticket #144 (Automatically scale variables in a dataset) created by
- In combination with #143
- 15:22 Ticket #143 (Exporter from HL dataset to libsvm data-format) created by
05/08/08:
- 08:50 Ticket #142 (The number of stored BadChildren can be reduced) created by
- I think it is not necessary to keep all bad children generated in the …
05/07/08:
- 21:29 Changeset [227] by
-
fixed evaluation of variables (returns NaN if index points out of the dataset) (ticket #139)
- 10:52 Changeset [226] by
-
fixed compilation problem introduced with: r225
- 01:39 Changeset [225] by
-
deleted obsolete class FunctionTree (ticket #139)
- 01:13 Changeset [224] by
-
implemented naive straight-forward way of serialization for baked-trees as tree-structure in XML. It should be possible to serialize and deserialize with the more compact linear representation as well. (ticket #139)
- 00:45 Changeset [223] by
-
moved evaluation code out of BakedFunctionTree into static class BakedTreeEvaluator to simplify code (ticket #139)
- 00:07 Changeset [222] by
-
fixed waiting for results (waithandle.waitall()) works only for maximally 64 wait-handles. So if execute more than 64 jobs in parallel we have to wait for each wait-handle separately.
- 00:06 Changeset [221] by
-
made the expiration time shorter to decrease probability to raise OutOfMemoryExceptions (quick fix for pspitzli meta-gp engines)
- 00:02 Changeset [220] by
-
worked on #139:
- fixed display of trees in the gui
- split list representation of tree into two lists code and data
- implemented static evaluation for all predefined functions (except ProgrammableFunction)
05/06/08:
- 16:16 Ticket #141 (Use WeakReference to store results in the EngineStore instead of ...) created by
- 13:21 Ticket #140 (ServerForm shows incorrect numbers for "Running jobs" and "Waiting Results") created by
- 13:00 Ticket #139 (Implement fast evaluator for GP function-trees) created by
- Some functions like addition, multiplication... can be statically …
- 12:46 Ticket #138 (Starting multiple applications concurrently causes some of the ...) created by
- Seems there is a synchronization problem in the splash-screen class.
- 12:27 Ticket #136 (Jobs that were not executed correctly are lost and the ...) closed by
- fixed: * Implemented restarting jobs in r129. * Also improved exception …
- 12:26 Ticket #137 (Stopping the GridClient while an engine is running throws an exception.) created by
- The channel is closed in the event-handler for the stop button. When a …
- 12:22 Changeset [219] by
-
- extracted communication code out of the DistributedEngine into class JobManager
- implemented a method to retrieve the JobState for a given job (specified by it's guid) in GridServer
- implemented restarting of jobs in JobManager
- improved exception handling in JobManager and DistributedEngine
(ticket #136)
- 09:16 Ticket #136 (Jobs that were not executed correctly are lost and the ...) created by
- DistributedEngine should handle exceptions more gracefully and …
05/04/08:
- 07:25 Ticket #134 (Wrong build time in XML fragments when publishing a plugin) closed by
- worksforme: Closing this issue as 'worksforme' after discussion with swagner.
