aMule Forum
English => en_Bugs => Topic started by: Underpass on May 13, 2012, 01:50:38 PM
-
Hello,
Yesterday I've compiled SVN version 10785 and noticed that problem. So I compiled the last version, 10787 and found the same problem.
So I switched back to SVN 10771 and it runs smoothly - moreover version 10771 seems to have a smaller CPU load.
If you need further information, please tell me how I can be of more help.
Thanks for your work. :)
-
Uh-oh.
There were some massive "code-cleanups" in versions 10782 - 10784. Apparently something went wrong there.
I can reproduce the crash.
I'm sure GonoszTopi will look into it as soon as he can.
-
Oh...
Yesterday when I was reading this I tried it, but nothing happened.
Right now it crashed here as well :/
-
Those "code-cleanups" revealed a real bug. Fixed in rev. 10788.
-
Hey, thanks for those code cleanups - and this bugfix :)
Some folk say that static code analyzers generate too many false positives - one response is that if they can't understand the code, then it probably needs rewriting to be understandable by humans.
I'll just say that finding that memory leak was worth the cleanup.
-
Well, we have found an easy-to-detect crash bug now.
I have no idea how many rarely-trodden code paths were touched as well and might now break. No problem about the ++it changes, but I'm a bit suspicious about the pointer casts. (Not to mention that the extra code required for the dynamic casts by far exceeds the benefit from the ++it optimization.)
-
and a memory leak.
Amule here has a 22 day uptime and is still only using a few hundred Mb.
Previously it had to be restarted every 5-6 days because it'd bloat out to a few Gb.