[patch-rfc] N2179: Exception Propagation in C++

Paolo Carlini paolo.carlini@oracle.com
Thu Aug 21 14:03:00 GMT 2008


Paolo Carlini wrote:
> I think it's correct, because at the same time I'm using 
> __exchange_and_add_dispatch, which essentially is a 
> __sync_fetch_and_add (*not* a |__sync_add_and_fetch as in your initial 
> patch)|. I also run your testcases through valgrind.
Gosh, I checked again and now I see some leakages, should have done 
something wrong last time...

However, the way to go certainly isn't initializining to 1, given that 
we have to use the __sync_fetch_and_add version...

Paolo.



More information about the Libstdc++ mailing list