controlflowgraph plugin moved to kdereview ...

[Sandro Andrade]

On Sat, Jun 19, 2010 at 8:45 AM, Milian Wolff <mail at milianw.de> wrote:
> On Saturday 19 June 2010 13:39:30 Sandro Andrade wrote:
>> Hi Niko,
>>
>> Thanks for reporting this, I actually removed some unnecessary DUChain
>> locks some weeks ago, may be I am
>> misunderstanding this.
>>
>> The function in which the assert hapenned
>> (DUChainControlFlow::processFunctionCall) is only called from other
>> functions which already holds a DUChain lock. I am considering the
>> locks are kept in ordinary function invocations
>> and when using signals/slots.
>>
>> I mean:
>>
>> X::m1()
>> {
>> DUChainReadLocker lock(DUChain::lock());
>> DUChainControlFlow::processFunctionCall();
>> }
>>
>> or
>>
>> X::m2()
>> {
>> DUChainReadLocker lock(DUChain::lock());
>> emit signalConnectedToProcessFunctionCall();
>> }
>>
>> Since connections aren't queued wouldn't the locks be kept in this both
>> cases ?
>
> You shouldn't hold across signals.

Fixed ! I've fully checked such cases, it should work now.

I'm wondering why I haven't faced such assert. It cames from incoming
arcs (uses) calculation and I'm using such feature quite often !

Thanks,
Sandro

> --
> Milian Wolff
> mail at milianw.de
> http://milianw.de
>