This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "MoleCuilder". The branch, ThreadFixes has been created at 38074bf6c963bbf4bc326b7725ba73b8eccd79f4 (commit) - Log ----------------------------------------------------------------- commit 38074bf6c963bbf4bc326b7725ba73b8eccd79f4 Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Fri Aug 29 11:46:02 2014 +0200 THREADFIX: Fixed ActionQueue with respect to mutex and threads. - run_thread is now last member variable, hence started after all other variables are initialized. - separated accesses to queues and to run_thread_isIdle flag. Also renamed its mutex for easier association with it. commit b4d77c0ac2e204573b6a9ae78bdbeddbfca1dd96 Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Fri Aug 29 11:43:10 2014 +0200 DOCU: Update on test policy with respect to now working valgrind tests. commit f825283d8fd2fad85e199cb3a32f98980f29ceb5 Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Fri Aug 29 11:44:20 2014 +0200 MEMFIX: ActionQueue was not stopped() when argument parsing failed. - this just caused an exception and the ActionQueue thread would be aborted uncleanly. This is only affected --enable-action-thread. commit 052c10a6e270eaffbd69149bcd5c6bbaadab9752 Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Thu Aug 28 10:48:24 2014 +0200 MEMFIX: Tesselation::operator() did not free linkedlist. commit 7ee21dd85ecc96fef7cf8272cb7e5755f57d1e65 Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Thu Aug 28 09:24:18 2014 +0200 MEMFIX: Many analysis_correlation function allocated return map twice. - outmap was the culprit, probably when empty (but allocated) was returned when no atoms are allocated, two new's appeared. commit 6b3e5eabbfb866e42ce576818aeadd95719a1ac6 Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Wed Aug 27 22:10:38 2014 +0200 MEMFIX: PartialNucleiChargeFitter lost memory. - operator() and constructMatrix() use heap memory to avoid always allocating memory for charge grids but did not always free'd the memory or allocated onto pointer with allocated memory. commit 96c7d2359255253121a4625cd3a9933834b44391 Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Wed Aug 27 21:59:01 2014 +0200 MEMFIX: PotentialRegistry's dstor forgot cleanup(). - also we purge static potential factory and registry instances in cleanup. - dstor of SerializablePotential was not virtual. commit 185e6f9e58b5418b9136e86a172d2cfb6627a2d7 Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Tue Aug 26 23:35:29 2014 +0200 MEMFIX: SaturationDistanceMaximizer::operator() did not delete Advocate. commit b0b2d2c9d316c1a252e1177412399808f116a2d9 Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Tue Aug 26 23:07:40 2014 +0200 MEMFIX: RealSpaceMatrix's and MatrixContent's serialization leaked memory. - serialization always works on fully instantiated RealSpaceMatrix, hence neither createViews nor re-creating content ptr instance is necessary. - MatrixContent newly allocated content but for RealSpaceMatrix it was still present as matrix dimensions have been set. Now it is freed when free_content_on_exit is true. - similar fixes to VectorContent. commit 7f1a1a7e897d84470656f52ade76a74fef5da063 Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Tue Aug 26 17:28:34 2014 +0200 MEMFIX: ActionQueue would loose memory when action failed. - we cleared the queues and forget to delete contained Action instances. commit 3d70e30645becb764684201ccfc596c838a74c3a Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Tue Aug 26 12:48:28 2014 +0200 MEMFIX: Mismatched delete in PdbParser::Save(). commit 09f615ff03d6c900818e07a53ee2c183bbf3bf47 Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Tue Aug 26 10:50:02 2014 +0200 MEMFIX: World's homologies not deleted. - main() now uses int status for storing saveAll() flag. commit 62c378a69a5535affe3a0039648a1cea69f07da3 Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Tue Aug 26 19:21:49 2014 +0200 FIX: Made checks in random number engine and distribution regression tests more robust. - with valgrind we see the command line with "lagged_fibbonacci.." and "uniform_smallint", too, resulting in more hits for grep. commit 5200294b51370a326ff7b6b894366ab63dd343f1 Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Tue Aug 26 19:12:05 2014 +0200 FIX: regression test add atoms outside boundary had given boundary conditions wrong in arguments. - TESTFIX: test is now marked XFAIL, before it failed because of wrong conditions. commit 25aa5ad00fe2dc8d803d1602be52c8f6e3bc16c7 Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Tue Aug 26 20:56:38 2014 +0200 FIX: Box::setM() did not use const ref as argument. commit 707a2b0d381b69df9ebf6bc57ce9ffe12636248e Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Wed Aug 27 21:58:28 2014 +0200 FIX: FitPotentialAction's getFirstGraphwithSpecifiedElements() access key_end(). commit 8946a0e70603b6a8930297686e4c9938010b2d9d Author: Frederik Heber <heber@ins.uni-bonn.de> Date: Wed Aug 27 20:52:44 2014 +0200 FIX: Process' dstor would always stop() even if not start()ed. - this caused assert failure from Observer cause it had not been signed, but was suddenly signed off. ----------------------------------------------------------------------- hooks/post-receive -- MoleCuilder