On Wed, 8 Sep 2010 13:28:36 +0100 John Sullivan <[email protected]> wrote:
> Originally submitted with a patch nearly two years ago, this appears > to have languished. As far as I can tell nothing else has fixed the > bug in the meantime. > > Could someone with check-in rights take a look at this: I think both > the problem and solution are straightforward. You submitted it under APR, but presented an HTTPD bug. That makes it less likely to get noticed (or taken seriously). There are a couple of issues. One: cross-platform issues: your patch is in /unix/ but references OS2. That flashes up "can't test so leave it" in me, and no doubt in other *X-oriented folks. Only on replying did I look carefully enough to figure out it doesn't in fact affect OS2. Two: the problem description is using httpd mod_cgi with a threaded MPM. This is a configuration that is explicitly not recommended: mod_cgid (or fastcgi) are documented as preferable. Easy to miss the underlying issue you're reporting! Three: apr is a library, and HTTPD is just one user among many. A change that might have side-effects on something else belongs in trunk only. Having said all that, bug me if I don't get around to reviewing it and either committing to trunk or raising objections in bugzilla. -- Nick Kew
