i'm using the batchbuild feature regularly, since a few revisions CB crashes when trying to batchbuild a workspace
here the commandline, which i use in a batchfile, which was working since half a year or longer ...
"D:\Programme\CodeBlocks\codeblocks.exe" -na -nd -ns -p=batchbld --build "D:\Devel\CodeBlocks\trunk\src\.my.CodeBlocks.workspace"
CB svn rev 2499, wx-widgets 2.6.3 Windows, unicode
and here the RPT file attached
[attachment deleted by admin]
here a call-stack trace,
might be better than the previous crash report.
[attachment deleted by admin]
Mmmmh... it's the bold line in the following code (sdk/messagemanager.cpp:443):
if (dlg->m_pText)
{
dlg->m_pText->AppendText(tmp + _T('\n')); // log to build log window
dlg->m_pText->ScrollLines(-1);
Manager::ProcessPendingEvents();
}
I don't see an issue here since dlg->m_pText is veriffeied before...?! Could you have a look if it points to the same construct the sources you've used?
i couldn't analyse the problem till now, but since my last installed "good" version, where everything was working ok for me, was rev 2472,
it has to do something with any modifications that happened later.
obviously the crash has something to do with the logging feature,
one of the suspect changes is in rev 2482, the commit descriptions says :
" * Cleaned up some logging code in compiler plugin. "
Quote from: MortenMacFly on May 24, 2006, 10:15:01 PM
I don't see an issue here since dlg->m_pText is veriffeied before...
it happened more than once in history of menkind, that a pointer isn't set to zero
and isn't valid any more at the same time ...
Quote from: tiwag on May 24, 2006, 10:27:04 PM
it happened more than once in history of menkind, that a pointer isn't set to zero
and isn't valid any more at the same time ...
Yes, this could be, of course! I did not think about that one. I'm looking for references to that pointer, maybe I can find a delete or similar without setting to zero...
Quote from: tiwag on May 24, 2006, 10:27:04 PM
Quote from: MortenMacFly on May 24, 2006, 10:15:01 PM
I don't see an issue here since dlg->m_pText is veriffeied before...
it happened more than once in history of menkind, that a pointer isn't set to zero
and isn't valid any more at the same time ...
The pointer is valid alright.
What I noticed though is that it works flawlessly when batch-building a project. It crashes only when batch-building a workspace. I believe this is enough of a clue to find out what's wrong ;)
Quote from: mandrav on May 24, 2006, 10:52:34 PM
I believe this is enough of a clue to find out what's wrong ;)
Could you have a look here http://forums.next.codeblocks.org/index.php?topic=3192.msg25284#msg25284 (http://forums.next.codeblocks.org/index.php?topic=3192.msg25284#msg25284), too? This could be related...?!
With regards, Morten.
Edit: Wrong URL in the fiurst place.
Quote from: MortenMacFly on May 24, 2006, 11:03:59 PM
Quote from: mandrav on May 24, 2006, 10:52:34 PM
I believe this is enough of a clue to find out what's wrong ;)
Could you have a look here http://forums.next.codeblocks.org/index.php?topic=3192.msg25284#msg25284 (http://forums.next.codeblocks.org/index.php?topic=3192.msg25284#msg25284), too? This could be related...?!
With regards, Morten.
Edit: Wrong URL in the fiurst place.
No, not related. Workspaces contain projects so it's not the same bug. Actually this is a strange bug and I 'm looking into it...