GNOME Bugzilla – Bug 509536
Leak in CalUtil.ICalToCalComponentArray
Last modified: 2008-02-26 10:43:20 UTC
Please describe the problem: Using the test program TestCal.cs, there is posssibly a leak in CalUtil.ICalToCalComponentArray. Steps to reproduce: 1. Add a webcal calender to evolution with rapid sync rate (e.g. 2/3 minutes). It will reproduce the bug quicker. 2. Compile and run TestCal. Observe that memory RSS keep increasing. 3. Comment out the body of the callback methods that use the above function and repeat the test. Memory stabilizes after some point. In my testing (with 4 webcal source), 20 mins saw more than 5 MB increase of RSS. Actual results: Expected results: Does this happen every time? Other information:
Up-ing the priority since the leak rate is pretty high and applications using evo-sharp are badly affected.
This is similar to the problem fixed in http://svn.gnome.org/viewvc/evolution-sharp?view=revision&revision=64 A memory leak when creating an array of CalComponent objects from icalcomponent and ECalComponent objects.
Created attachment 105691 [details] [review] Proposed Patch Patch releases e_cal_component. Would like to get this tested ( runs good with TestCal.exe. ) and commit it in trunk asap.
Hmm... testing this (i.e. building evo-sharp from source) is going to be difficult :-( If you build them and share the glue .so and the .dll file, I can try to put them in the right place and see if that works. On the other hand, if testcal worked, then the problem should be fixed.
I was able to test this patch, thanks to Johnny and it works perfectly.
Thanks a lot dBera for the test run. Patch committed to trunk svn rev 161. Closing bug.