On Tuesday, 22 November 2011 21:27:08 Philip M. Gollucci wrote: > 1) use dist.apache.org for releases and distribution > I'll handle this and update RELEASE accordingly. > 2) drop support for perl < 5.10.0 [patch ready]
+1 > 3) drop support for httpd 2.0.x and lower [patch ready] +1 > 4) stop waiting and merge branches/httpd24 and branches/threading to > trunk (Thanks Torsten!) Not yet ready. It compiles and even works somehow. But all the new stuff in 2.4 is not ported yet, particularly the new per-module logging and the AAA stuff. Here we need to decide whether we want to implement the old perl side interface based on the new httpd interface or if it is better to pass the new interface through and drop the old interface for httpd 2.4. I think I'd prefer the second even though it means rewrite of a lot of tests and modules out there. As for threading, I think a rewrite with a slightly changed interface would be good: 1) take the refcount idea from the threading branch to attach interpreters 2) disconnect interpreter pools from vhosts: <PerlInterpPool Pool1> PerlSwitches ... </PerlInterpPool> <VHost ...> PerlUseInterpPool Pool1 </VHost> <Location ...> PerlUseInterpPool Pool2 </Location> 3) stop using multiple pointers to reference in principle the same thing. Currently the perl interpreter assinged to a thread is stored in a thread specific variable (PERL_SET_CONTEXT). Perhaps we can store our interpreter pool element in a variable in that interpreter. Thus, the interpreter knows everything to put itself back to the pool and we do not need lists to maintain interpreters in use. Instead each pool is simply an apr_queue_t of free interpreters. I am currently in a project that eats up all my time. So, don't expect much help from me till end of the year. > 5) Stop versioning APR/APU 0.9, and start supporting apr where > apr/apr-util are the same thing > 6) If you missed it httpd/apreq/trunk C bindings are not part of httpd > for 2.5.x+. The perl bindings need a new home. For the moment, > mod_perl is the only logical choice, so they will be relocated and > apreq will be 'closed down' I'd add 7) Have a look at the futex problem recently reported by Max Barry. 8) sort out the large file support reported by Marco Walther in the thread "[mp 2.0.5] Early core dump:-( OpenSolaris-x86, Apache 2.2.21 & Perl 5.14.1" Torsten Förtsch -- Need professional modperl support? Hire me! (http://foertsch.name) Like fantasy? http://kabatinte.net --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@perl.apache.org For additional commands, e-mail: dev-h...@perl.apache.org
