Re: Bugfix: Crash in PacketSession::addSession()

Hi Atul,

You are right we should always think about what we can learn after every new thing we create or bug we fix. That helps our own understanding and gives a starting point for other people, so they don’t have to reinvent the wheel. These people would need to read some docs though ;-). So I plan to write some new section to the docs regarding problems related to races when deleting QObject derived classes. I think it would be nice to have a docs chapter only about programming techniques.

Good tools can really make life easier! I didn’t know about HelGrind. Just checked their web page and it says that that HelGrind does not work with Valgrind 3.1.0. Some years ago I had used Boundschecker and that was a really great tool! It helped me find memory problems quite easily. I guess the latest version would be even better.

What I really want to have is a logging framework which can log per object events and then maybe have a tool which will help us inspect the life cycle of the object where the crash occurs. If you know what exactly happened when, then it is so much easier to understand the cause of the problem and fix it. What the debugger tells you is there the software breaks, not why it broke. I have searched for a logging framework and found log4cxx, log4c and log4cpp. I’ll try them out and let you know. I’ll be glad if someone else is also interested in playing around with this stuff and if we can exchange experiences.

Regards,

Peter

Would you like to post a relpy?


This post is a reply to:
Re: Bugfix: Crash in PacketSession::addSession()
Hi Peter Great job mate... Now that the bug is uncovered, I like to think  "How fast we could have resolved such bug?" These days, people are talking a lot about coverity (more...)

No follow-ups yet.