Dale Ghent wrote: > > > My issue with including libpthread on Solaris builds that do not use > HAVE_PTHREAD_SERIALIZED_ACCEPT is that, as stated by the Sun docs at the > URL I posted yesterday, that unnecessary overhead is introduced into the > proces as thread-related structures and environment is set up and > maintained for the process... which is then never utilized. > > It seems like a performance hit we can avoid. We can stick -lpthread in > LIBS and then enable HAVE_PTHREAD_SERIALIZED_ACCEPT for the default. > Otherwise, having -lpthread in there and NOT using > HAVE_PTHREAD_SERIALIZED_ACCEPT would just be a performance handicap for > 90% of Solaris builds out there that do not enable > HAVE_PTHREAD_SERIALIZED_ACCEPT by hand. >
Because we have HPSA added in ap_config.h for Solaris, it *is* enabled by default (we allow pthread, sysvsem and fcntl). The actual mutex method is runtime configurable, but if we want to allow pthread as one of the options, we *have* to compile against lpthread. By initial thought was have Configure smart enough to do a test compile under Solaris and see if HPSA is defined, and if so, add in in pthread lib. But after thinking it over, if someone is smart enough to want to totally disable even the prospect of using pthread mutex on Solaris, and edit ap_config.h to remove that, then they should also be smart enough to notice that lpthread needs to be removed as well. Maybe we should do this anyway, "just in case" but I don't think it's a show stopper... heck, some minor comments in the docs and code would be enough. -- =========================================================================== Jim Jagielski [|] [EMAIL PROTECTED] [|] http://www.jaguNET.com/ "A society that will trade a little liberty for a little order will lose both and deserve neither" - T.Jefferson
