I've been looking at the failure Randy reported with t/api/add_config failing when run multiple times.
Took a while, but I finally got my eyes open today. It's an issue of thread-safety that was completely overlooked. Trying to do something like: $r->add_config(['<Location /foo>',[...],'</Location>']); Is not, and cannot be thread-safe, as it needs to access the server's core config at one point, leading to all sort of nastyness. Another way to achieve the same thing is with: $r->server->add_config() At request-time. With preform, this will work, but obviously, only affect the current child process. In my opinion, this is way *wrong* and needs to be fixed. Similarly to how .htaccess files work, you should only be able to affect configuration of the current request when using add_config() at request time. It's been fixed/implemented in r386774. (It's not documented yet, patches welcome) Randy, I'd like to know if your test failure dissapears with this change. -------------------------------------------------------------------------------- Philippe M. Chiasson m/gozer\@(apache|cpan|ectoplasm)\.org/ GPG KeyID : 88C3A5A5 http://gozer.ectoplasm.org/ F9BF E0C2 480E 7680 1AE5 3631 CB32 A107 88C3A5A5
signature.asc
Description: OpenPGP digital signature
