Jim Segrave wrote: > On Tue 27 Feb 2007 (09:11 +0100), Massimiliano Maini wrote: >> Hmmm, strange. I thought Jon, wise man, had left all the old non-MT code >> in place and surrounded all the new MT code with #if USE_MULTITHREAD ... >> What exactly did you test ? Did you use an executable compiled with >> #define USE_MULTITHREAD 1 on a single proc PC or did you use executables >> compiled WITHOUT USE-MULTITHREAD ? > > My apologies for the confusion. It only applies to the CVS Main > branch, not the RELEASE_0_15 branch. But under Main, it didn't matter > if you have used --enable-threads or not, prior to the patch I checked > in this weekend. My patch fixes this problem, although I'm worried > about other consequences of rcRollout being a global variable in a > threaded environment. That's something that goes back years and the > number of places rcRollout is accessed was what discouraged me from > trying to address this when I did the rollout changes a couple of > years back.
Although it's (rcRollout) accessed in lots of places, it doesn't seem to get altered in the multi-threaded code (RolloutLoopMT). Removing all the globals would be nice, I think this will only happen slowly over time... Jon
signature.asc
Description: OpenPGP digital signature
_______________________________________________ Bug-gnubg mailing list [email protected] http://lists.gnu.org/mailman/listinfo/bug-gnubg
