Not totally lost, it depends on whether the failing output filter is before or after the http_header filter. In the latter case, we already sent (part of) the response to the client, so we must abort. But in the former case, ap_die() will generate a 500, so to not let the client without any response.
On Thu, Jun 4, 2015 at 7:20 PM, Eric Covener <cove...@gmail.com> wrote: > maybe part of what I am missing is that any output filter failure is a lost > cause, and the mapping happens on the input filter side. > > > On Thu, Jun 4, 2015 at 1:15 PM Eric Covener <cove...@gmail.com> wrote: >> >> I'm a little confused by AP_FILTER_ERROR as used in the handlers we have >> now. Is this return code evolved to mean only to issue a 500 error if the >> response has not been committed? >> >> /** Returned by any filter if the filter chain encounters an error >> * and has already dealt with the error response. >> */ >> #define AP_FILTER_ERROR -102 >> >> But now the handlers have this pattern: >> status = ap_pass_brigade(r->output_filters, bbout); >> if (status != APR_SUCCESS) { >> /* no way to know what type of error occurred */ >> ap_log_rerror(APLOG_MARK, APLOG_DEBUG, status, r, >> APLOGNO(01410) >> "reflector_handler: ap_pass_brigade returned >> %i", >> status); >> return AP_FILTER_ERROR; >> } >> >> So we don't actually retain whether some filter really handled the error. >> I think in practice this is okay, because the handlers can't make a very >> smart decision here anyway. >> >> On Thu, Jun 4, 2015 at 12:44 PM William A Rowe Jr <wr...@rowe-clan.net> >> wrote: >>> >>> More context at your fingertips without refreshing httpd-2.2 branch, >>> first... >>> >>> https://bz.apache.org/bugzilla/show_bug.cgi?id=57832 >>> >>> On Thu, Jun 4, 2015 at 11:26 AM, William A Rowe Jr <wr...@rowe-clan.net> >>> wrote: >>>> >>>> [Changing subject, don't mean to hijack the 2.4 activity train] >>>> >>>> There is a modestly important patch, already backported to 2.4.x branch, >>>> that is still sitting in 2.2 status. Could one more committer please >>>> review >>>> and vote on that remaining fix? >>>> >>>> Because it helps to avert an unintended doubled response in some edge >>>> cases, I consider this one important enough to hold up 2.2 tag for some >>>> more hours. >>>> >>>> Bill >>>> >>>> >>>> On Tue, Jun 2, 2015 at 4:36 PM, William A Rowe Jr <wr...@rowe-clan.net> >>>> wrote: >>>>> >>>>> On Tue, Jun 2, 2015 at 6:32 AM, Jim Jagielski <j...@jagunet.com> wrote: >>>>>> >>>>>> Although there are some cool things that I'd like to see in >>>>>> 2.4.13, I don't want to hold off any longer (plus, those >>>>>> cool things would be good incentive for a 2.4.14 sooner >>>>>> rather than later). >>>>>> >>>>>> I plan to T&R 2.4.13 on Thurs, by Noon eastern. >>>>> >>>>> >>>>> +1, planning to match you with a T&R of 2.2.30 on the same timetable. >>>>> >>>>> There is a nominally important last patch in 2.2 STATUS, if a third >>>>> pair of eyes have the cycles to review it. >>>> >>>> >>> >