|
|
Threads-Devel : |
From: Anthony Williams (anthony_at_[hidden])
Date: 2006-03-23 06:30:00
Roland Schwarz <roland.schwarz_at_[hidden]> writes:
> Anthony Williams wrote:
>> >From the logs you posted, it seems the problem was a using declaration I added
>> to boost::mutex. This was redundant, so I've removed it, and hopefully all the
>> problems will go away.
I spoke to soon on that one. I'll have to dig out my copy of VC6 to get this
working.
>> The once failures are concerning, as I thought my implementation of that was
>> pretty tidy.
>>
> I added the free borland compiler, but it seems that it also is
> deprecated by boost. While I do not know the
> exact reasons I think this is unfortunate since it is the current freely
> available free compiler form the borland
> website.
There was a lot of discussion about this on the main boost list recently. I
think it boiled down to the fact that this is an old compiler, and rather
non-conforming, so required lots of workarounds. This was rather controversial
with some, so it's probably best not to go into it here. Check the archives
for details.
> As you can see the once still does not work cleanly. Could it be that my
> setup is not correct?
I doubt it. I expect this issue (and the memory violations) is to do with ...
> I am running on a pentium with hyperthreading, don't know if this matters...
... this. A P4 with HT mimics a dual processor system in some ways, so will
expose more deadlocks and race conditions than a normal P4. This is a good
thing.
Anthony
-- Anthony Williams Software Developer Just Software Solutions Ltd http://www.justsoftwaresolutions.co.uk