337746f395c4bb401c832af0feaeca13cdc45d5a Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/9546bfef5a360e6da44e13e26cd13b98a574ae7a 9546bfef5a360e6da44e13e26cd13b98a574ae7a

2009-02-28T04:03:06-08:00

2009-02-28T03:58:05-08:00

quickfix release 0.7.3 This makes 0.7 run on activesupport. This is probably the last release of the 0.7 branch, will release 0.8 momentarily. 8c8b8a2c2985ad6cb22386283039a01f7a5a0536 Sylvain Joyeux sylvain.joyeux@m4x.org 3f5b1f3c04dd5ff0d08d6ab14927b1de8985b101 Sylvain Joyeux sylvain.joyeux@dfki.de http://github.com/doudou/roby/commit/337746f395c4bb401c832af0feaeca13cdc45d5a 337746f395c4bb401c832af0feaeca13cdc45d5a

2009-02-28T03:41:22-08:00

2009-01-06T06:30:45-08:00

core: adapt to the newest activesupport release This commit fixes a require path that has changed in the latest activesupport release. c81bf55b6d494bba91fd6bf59fe642b5112a080f Sylvain Joyeux sylvain.joyeux@m4x.org cff3e8af29bc8f7e4c43fa3f7b82407a06621466 Jan Steinmetz Jan.Steinmetz@dfki.de http://github.com/doudou/roby/commit/3f5b1f3c04dd5ff0d08d6ab14927b1de8985b101 3f5b1f3c04dd5ff0d08d6ab14927b1de8985b101

2008-09-16T07:36:38-07:00

2008-09-16T07:02:28-07:00

Added some more validation for emission calling of tasks. Added the methods check_emission_validity and check_call_validity in event.rb which are both called in call/emit and call/emit_without_propagation for further error handling. Extracted the exception refinement in task.rb for the EventNotExecutable exception. The check_*_validity methods were overwritten to make use of the refinement. Added tests in test_task.rb to test the raising of the EventNotExecutable exception. 69676033a7f2a429a4f1e58724ffb0eaeb1a877d Jan Steinmetz Jan.Steinmetz@dfki.de 069bb726aca8c63744ffb8909b445a5ed301b233 Sylvain Joyeux sylvain.joyeux@dfki.de http://github.com/doudou/roby/commit/cff3e8af29bc8f7e4c43fa3f7b82407a06621466 cff3e8af29bc8f7e4c43fa3f7b82407a06621466

2008-09-16T03:33:50-07:00

gui: be compatible with various internalPointer/internalId behaviour in QtRuby Various versions of QtRuby handle AbstractModelItem#internalPointer and internalId with various levels of bug-ness. Try to handle those various behaviours nicely so that roby-log works on various versions of QtRuby ffe3a32bc13c48ea891a3602377696c3908411d6 Sylvain Joyeux sylvain.joyeux@dfki.de a4a3271692e109697d171f0ccaf32d3879a00c81 Sylvain Joyeux sylvain.joyeux@dfki.de http://github.com/doudou/roby/commit/069bb726aca8c63744ffb8909b445a5ed301b233 069bb726aca8c63744ffb8909b445a5ed301b233

2008-09-16T03:30:46-07:00

dist: search for rbuic4 also in /usr/lib/kde4/bin The current Ubuntu version places rbuic4 there, so we have to search for it. 7f33f69c43322440f09899b4bf7f12c51e0e95bd Sylvain Joyeux sylvain.joyeux@dfki.de 0268b55af5fdb04927f038bb840e80ef9fc43ed0 Sylvain Joyeux sylvain.joyeux@dfki.de http://github.com/doudou/roby/commit/a4a3271692e109697d171f0ccaf32d3879a00c81 a4a3271692e109697d171f0ccaf32d3879a00c81

2008-09-16T03:29:55-07:00

dist: completely ignore Hoe whenever an exception is raised from it The problem is that different versions of Hoe can depend on specific versions of RubyGems. We should therefore disable completely distribution if Hoe raises (for whatever reasons) 3122594aa2cdc029bc740a6a2799749a7d406b46 Sylvain Joyeux sylvain.joyeux@dfki.de 42656ace91001a66a1c4873014c1a7b1f45ad9e3 Sylvain Joyeux sylvain.joyeux@dfki.de http://github.com/doudou/roby/commit/0268b55af5fdb04927f038bb840e80ef9fc43ed0 0268b55af5fdb04927f038bb840e80ef9fc43ed0

2008-09-08T07:38:01-07:00

2008-09-08T07:30:45-07:00

app: set a default sensible configuration for IRB in scripts/shell To make the shell usable, the INSPECT_MODE *must* be disabled. Moreover, the default shell prompt is really ugly. Therefore, scripts/shell sets a Roby shell prompt, which is not configurable for now. 9201711474df89af15aa2d34406ecf13a29f72bb Sylvain Joyeux sylvain.joyeux@dfki.de 8c380d0d291fc8058d82d055a623dc9b2497606a Sylvain Joyeux sylvain.joyeux@dfki.de http://github.com/doudou/roby/commit/42656ace91001a66a1c4873014c1a7b1f45ad9e3 42656ace91001a66a1c4873014c1a7b1f45ad9e3

2008-09-08T07:36:29-07:00

2008-09-08T07:34:29-07:00

app: display Roby's version at startup This is mainly for debugging purposes when multiple versions of Roby are lying around. ca1e0139496f46ccb85e656a79e149c9c657b946 Sylvain Joyeux sylvain.joyeux@dfki.de 6d645f042658fe7cace252ce50a5300e992a9268 Sylvain Joyeux sylvain.joyeux@dfki.de http://github.com/doudou/roby/commit/8c380d0d291fc8058d82d055a623dc9b2497606a 8c380d0d291fc8058d82d055a623dc9b2497606a

2008-09-08T07:36:26-07:00

2008-09-08T05:57:18-07:00

log: properly handle the case where no display has been defined Until now, the whole program was crashing. It now ignores the user's attempt to move around in the log file. The proper fix would be to disable the motion controls until one display has been defined. 38f2634f92de0867f743a580d30469215c30c418 Sylvain Joyeux sylvain.joyeux@dfki.de c750dcb589efc9ed2afb34a6636e993266b29a8b Sylvain Joyeux sylvain.joyeux@dfki.de http://github.com/doudou/roby/commit/6d645f042658fe7cace252ce50a5300e992a9268 6d645f042658fe7cace252ce50a5300e992a9268

2008-09-08T07:36:20-07:00

2008-09-08T05:48:09-07:00

handle wrong or incomplete index files The first time the index file is loaded, the index data is checked agaoinst the read log file by loading a 100 cycles that are registered in the index files. If that check does not pass, the index file is rebuilt. This will make sure that a corrupted index file will be rebuilt. Moreover, it will also rebuild an index file which is not the one of the current log file (i.e. dowloaded a new event log with an index file lying around) 279975cc96881703060b0892d7267635f47d0a9d Sylvain Joyeux sylvain.joyeux@dfki.de 7002be4e9a1a482343d7bac46ef32e22c23d3238 Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/c750dcb589efc9ed2afb34a6636e993266b29a8b c750dcb589efc9ed2afb34a6636e993266b29a8b

2008-06-14T09:56:03-07:00

2008-06-14T09:53:47-07:00

[dist] fix Hoe's description There's really no need to put all of the README in the description. Limit it to the purpose of Roby, and add a line "for more information" 6fe05df708dd0a1141d47b027be2c96dc8401613 Sylvain Joyeux sylvain.joyeux@m4x.org df22f55a88fd2ae0fb1bfe2e30adcb35826acde8 Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/7002be4e9a1a482343d7bac46ef32e22c23d3238 7002be4e9a1a482343d7bac46ef32e22c23d3238

2008-06-14T09:41:59-07:00

2008-06-14T09:41:36-07:00

[core] be compatible with 1.8.7 b28a579a83a293a0fbb5ed9613d483ec8a8d2f73 Sylvain Joyeux sylvain.joyeux@m4x.org b2d80456a60c758314f5a0fb2da6b68435082151 Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/df22f55a88fd2ae0fb1bfe2e30adcb35826acde8 df22f55a88fd2ae0fb1bfe2e30adcb35826acde8

2008-06-14T09:41:21-07:00

[dist] refuse loading on 1.9 b9fafc954b5825505169c73ad4a699b65a6209a5 Sylvain Joyeux sylvain.joyeux@m4x.org 6412d4d78d020da62de47350cc2d6ee2a154f59b Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/b2d80456a60c758314f5a0fb2da6b68435082151 b2d80456a60c758314f5a0fb2da6b68435082151

2008-06-14T09:40:51-07:00

[dist] utilrb 1.3 was a bad release, require 1.3.1 fb621dec95cb4742511b2b37cb272f43114c87c5 Sylvain Joyeux sylvain.joyeux@m4x.org 4bf13ce196c18d4be21b53174d01f9e935434211 Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/6412d4d78d020da62de47350cc2d6ee2a154f59b 6412d4d78d020da62de47350cc2d6ee2a154f59b

2008-06-14T08:12:38-07:00

[core] prepare release of 0.7.2 f77989549c877fe814f7b607d13abcb36e7af8b7 Sylvain Joyeux sylvain.joyeux@m4x.org cc7027bfe1ff97979cf492bd6bde309815ad421a Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/4bf13ce196c18d4be21b53174d01f9e935434211 4bf13ce196c18d4be21b53174d01f9e935434211

2008-06-06T05:15:19-07:00

[doc] update 7263a936f6e2001a8558c7c64de5e454b1edf0ac Sylvain Joyeux sylvain.joyeux@m4x.org 5b73146c8de8eaf1b445c558254a08cd3bf43add Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/cc7027bfe1ff97979cf492bd6bde309815ad421a cc7027bfe1ff97979cf492bd6bde309815ad421a

2008-06-06T03:28:50-07:00

[distributed] more robust ConnectionSpace#receive For now, this method did not expect socket reading to fail. This patch fixes that, by catching possible exceptions which would come from Socket#read 0d14f667df2642800af43ccfe6f5a596f1946d81 Sylvain Joyeux sylvain.joyeux@m4x.org 7df1db428ed18d0d76a723fe1b9428647cdfbce5 Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/5b73146c8de8eaf1b445c558254a08cd3bf43add 5b73146c8de8eaf1b445c558254a08cd3bf43add

2008-06-06T03:06:45-07:00

[dist] now depend on Utilrb 1.3 for Utilrb::WeakRef c783421d3aa5c61350e929d2fa0db6b438b30eaa Sylvain Joyeux sylvain.joyeux@m4x.org c276afe6ff2c8cfdc570a2facd8196696454e5f4 Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/7df1db428ed18d0d76a723fe1b9428647cdfbce5 7df1db428ed18d0d76a723fe1b9428647cdfbce5

2008-06-06T03:06:02-07:00

[core] remove wrong assertion in Control#run There was an assertion that @thread is non-nil at the end of #run. It is false, as it is perfectly possible that the thread starts and ends before the main thread is called again. Conflicts: lib/roby/control.rb 0b05d3340e43f00cb90ebfacbcf4e14b69652f52 Sylvain Joyeux sylvain.joyeux@m4x.org 7fd1a22aa9a1ed91d687d9c505076d5a34942760 Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/c276afe6ff2c8cfdc570a2facd8196696454e5f4 c276afe6ff2c8cfdc570a2facd8196696454e5f4

2008-06-06T03:03:26-07:00

[core] use Utilrb::WeakRef to remove the impact of the standard ::WeakRef The standard WeakRef implementation is Ruby-only and subclasses Delegate, which is way too costly for our use. This patch uses the C++-implemented Utilrb::WeakRef, which is much leaner (and finally fixes the memory leak) Conflicts: lib/roby.rb 5f50fa45efe0d73e8cb883fc18f31f74a7c0b4a1 Sylvain Joyeux sylvain.joyeux@m4x.org 3b91f1f5b4a76e1d71bed0131de21a1b6fd3294a Sylvain Joyeux sylvain.joyeux@dfki.de http://github.com/doudou/roby/commit/7fd1a22aa9a1ed91d687d9c505076d5a34942760 7fd1a22aa9a1ed91d687d9c505076d5a34942760

2008-06-04T09:43:46-07:00

2008-06-04T09:27:55-07:00

[core] fix Event#sources Test suites are there for something ... I suck. This problem has been introduced in 926ab1e83cf86a395e468448be19a534fc72a0f6. 765837078b0a6e92b456d0909a54391aff6b90cd Sylvain Joyeux sylvain.joyeux@dfki.de 0582d68f68cd8f3d9704d9c1997267ff9a183eb7 Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/3b91f1f5b4a76e1d71bed0131de21a1b6fd3294a 3b91f1f5b4a76e1d71bed0131de21a1b6fd3294a

2008-06-04T09:30:11-07:00

2008-06-01T13:28:47-07:00

[core] fix memory leak in Event Until now, Event#source was keeping around the sources of propagation. Since the sources themselves were keeping their sources -- and since Event keeps references on generator and tasks -- it was potentially a source of memory leak for long-running high-level tasks. This patch references the event sources by using WeakRef. That should not be a problem since #sources is not used often (mainly in error handling) 27b6e4bcf7b7f4332ba61963b429ef19297a8421 Sylvain Joyeux sylvain.joyeux@dfki.de 902ea8b792caaf9f356ec9c0526a253c5e6ad097 Sylvain Joyeux sylvain.joyeux@dfki.de http://github.com/doudou/roby/commit/0582d68f68cd8f3d9704d9c1997267ff9a183eb7 0582d68f68cd8f3d9704d9c1997267ff9a183eb7

2008-06-04T08:58:24-07:00

2008-06-04T08:58:10-07:00

[core] fix bug and memory leak related to EventGenerator.gather_events .gather_events has been designed to have the propagation code push the set of fired events in collections, so that plan analysis code can avoid iterating on the whole plan to find the interesting events. Until now, the automatic de-registration of the events when they become unreachable was actually not done, resulting in a memory leak. This commit fixes it -- and updates the test suite so that it does not appear anymore. f59be13983984a02774ec83c4796a9bda6277d11 Sylvain Joyeux sylvain.joyeux@dfki.de 93f5db56945e2b9e6661cb575eb0b82b64397cc2 Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/902ea8b792caaf9f356ec9c0526a253c5e6ad097 902ea8b792caaf9f356ec9c0526a253c5e6ad097

2008-05-29T14:38:56-07:00

[core] prepare to release v0.7.1 ad3dd505bb81e63b462fda2e0fc5f821f831e88d Sylvain Joyeux sylvain.joyeux@m4x.org f7e19a8a22800357fa16f0d5eb2c01747775f111 Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/93f5db56945e2b9e6661cb575eb0b82b64397cc2 93f5db56945e2b9e6661cb575eb0b82b64397cc2

2008-05-29T14:36:15-07:00

[dist] mention the BOOST_DIR envvar for rake setup ade5222aceaf717fc3ea65b8751759e5c37f4465 Sylvain Joyeux sylvain.joyeux@m4x.org 478442ff34d855aff254acf649f9bda4e61c7ca5 Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/f7e19a8a22800357fa16f0d5eb2c01747775f111 f7e19a8a22800357fa16f0d5eb2c01747775f111

2008-05-29T14:32:23-07:00

2008-05-29T14:07:47-07:00

[test,core] fix synchronization issues for operations w/o control thread A special mode of operation is available when no control thread is available, mainly for ROby's own testsuite. In particular, Roby.execute yield directly and event propagation is done directly in #emit and #call. Until now, those three operations where not synchronized. This introduced problems when a separate thread does plan modifications in a Roby.execute context while the testing thread do also some plan modifications. In particular, the move of #commit_transaction inside the plan thread is a source of such race conditions. In principle, the problem only shows up in Roby's own test suites, as normal operations are done with a separate execution thread. 2c47920c0c132808ba5e159eb131589b0490b8ed Sylvain Joyeux sylvain.joyeux@m4x.org 411fdf5e6dd81a6fffd78d7d26333c0eb92c74e8 Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/478442ff34d855aff254acf649f9bda4e61c7ca5 478442ff34d855aff254acf649f9bda4e61c7ca5

2008-05-29T14:32:22-07:00

2008-05-29T14:13:00-07:00

[git] update gitignore after the change of name for extensions 7ef328b51009376d23e082cf967491d6eebacf2a Sylvain Joyeux sylvain.joyeux@m4x.org f8d1d313e39a7d2712f36b5a5f9c7b6651847ba0 Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/411fdf5e6dd81a6fffd78d7d26333c0eb92c74e8 411fdf5e6dd81a6fffd78d7d26333c0eb92c74e8

2008-05-29T14:31:21-07:00

2008-05-29T13:30:24-07:00

[test] add test_thread_task to suite_core 9ec31f8b5c80eb4b752836bd5b29598c4570592f Sylvain Joyeux sylvain.joyeux@m4x.org 5ace95ce92a425063c3476e12ef3ca499c863f4e Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/f8d1d313e39a7d2712f36b5a5f9c7b6651847ba0 f8d1d313e39a7d2712f36b5a5f9c7b6651847ba0

2008-05-29T14:31:21-07:00

2008-05-29T14:10:21-07:00

[core] fix a potential race condition in PlanningTask The call to #commit_transaction and the setting of @result were done in an asynchronous way, in the planning thread. In practice, it could led for a short timespan where #planned_task returned nil instead of a meaningful value. This patch does the following: * it introduces a block parameter to #commit_transaction which, if given, is called after the commit but in the same Roby.execute context * @result is set in such a block 8118fa0820d25de5c92500ae6ed986fc3fcf8573 Sylvain Joyeux sylvain.joyeux@m4x.org aa854bfc8bdf36a8195aa6c73700f5a13f0fa6ad Sylvain Joyeux sylvain.joyeux@m4x.org http://github.com/doudou/roby/commit/5ace95ce92a425063c3476e12ef3ca499c863f4e 5ace95ce92a425063c3476e12ef3ca499c863f4e

2008-05-28T14:50:03-07:00

[test] remove timing dependency by using FlexMock's proxying facility Until now, some tests were relying on timing to check temporal behaviour. This is not reliable since execution times vary from computer to computer. Flexmock allows to temporarily change an object method. We use this facility here in one testsuite to wrap Time.now, therefore removing the timing problem. ae77d98e3f6a6e06a994a4732d30a9d1d5a635fa Sylvain Joyeux sylvain.joyeux@m4x.org