At 02:20 PM 3/27/2003, William A. Rowe, Jr. wrote: >[How did that happen??? Again, with a patch.]
Ok, regressions on win32... Starting sans-DSSL, so that the default ssl.conf is not loaded; [Thu Mar 27 14:48:16 2003] [warn] Init: Session Cache is not configured [hint: SSLSessionCache] Ick, now does that have to be a *warning*??? Either info or debug level should be sufficient! But we have no SSLMutex errors... now to SSLMutex default (with -DSSL), we get no startup errors at all. Of course as I point out in the prior message, each process is running in their own SSLMutex (one per generation) with no mutex protection during graceful restarts. So we can't support the .conf default for ssl.conf of file:logs/ssl_mutex, but I believe that's better hacked in awk to 'default' until the other patch is applied, and then move to default:logs/ssl_mutex as a works-everywhere solution. So my other two patches for ssl_engine_config and ssl_engine_mutex can be deferred for 2.0.46, and recognize that the cross-restart mutex has always been broken for all platforms, so this is not a regression. Bill
