Howdy, On Jan 8, 2008 10:43 AM, mabshoff <[EMAIL PROTECTED]> wrote:
> Ok, the culprit points to a cleanup function of m4ri, which under > normal conditions is called only once. I am not seeing the above issue > under valgrdind with "pure" Sage 2.10.alpha0, but as I just learned > yesterday problematic free() calls might be hidden by other bugs. > Imagine my surprise that once I fixed #1092 or #1093 some invalid free > popped up. We currently have a problem in tp_new (see #1337) which is > the only culprit on our end that could hide that. But the question > remains if IPython does something funny with threads and somehow > triggers more than one cleanup call. We could add some check for NULL > inside destroyAllCodes(). So, anybody with a better knowledge of > IPython: It seems that threads are involved, so could this somehow be > triggered? Sorry for the slow reply. Bit swamped 'round here... Yes, the issue is really that I explained myself poorly to Jaap. He's running import IPython IPython.Shell.IPShellWX().mainloop() Inside an already started Sage session. But at the exit of that mainloop(), ipython will tear down the threading support (absolutely necessary for WX to work without blocking the interactive console). That mainloop is hooked into the WX event loop, so it can't really be restarted. The solution is to have Sage start ipython with the wthread option, if you want full Sage support. As a starter, you can test this. Make a little file that's simply #!/path/to/sage/python import IPython IPython.Shell.IPShellWX().mainloop() make it executable, and run it *standalone* in a sage-sh configured shell. This standalone ipython is equivalent to running ipython -wthread and will simply exit at the end. If that works and closes without crashing, then someone more familiar than myself with how sage starts itself can then offer an option for sage to fire up ipython with '-wthread' at startup. This will ensure that the threads cleanup only happens when sage itself exits, not in the middle of the enclosing Sage session. My originally incomplete explanation led Jaap to have a main event loop inside another one. That's always bad news. I hope this is now clearer. Cheers, f --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/ -~----------~----~----~----~------~----~------~--~---
