Hi Kai, Thank you for your comments.
Kai Ahrens wrote:
in case of a crash, the stacktrace reporting tool should come up immediately, so that anybody getting a crash should be able to send crash related and relevant information (stacktraces) to us.
Yes. Just one thing. Investigating this kind of crash using a stack trace could be sometime hard. The surface of crash might look like a segment violation, but the real cause might be use of trickily, randomly, contaminated data overwritten by another thread or things like that. That is why i would like to propose a code review by those who have an experience in multi-thread programming. And from my few experiences of this kind of de-bugging, code reviewing was in many cases easier and more effective than attacking pieces of violently broken puzzle.
In case you have a stacktrace by hand, it would be great to send this to the list for evaluation. IMO it doesn't make any sense to dig around in code that has something to do with caching of different kind of objects. And yes, we do have a _lot_ of caches inside OOo for many good reasons.
I see. Anyway, I will try to reproduce a phenomenon. There is another headache. People also have reported another problem beyond a crash. In their problematic situation, OOo/SS seems working but refuses saving a document into a file for an error. I am still not sure what is the cause of such an error. In such a case, they have no practical way to report their situation and reluctantly restart or logout from their computer, loosing their documents.
Beside this particular issue, having code reviews on a a regular basis is always a good idea and already done by the (some) developers of the code, so that giving us a hint for the starting point of a next review (as you did) is also very valueable for us.
Right. Best regards, Tora --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
