On 12/17/2005 06:30 PM, Jim Jagielski wrote: > > On Dec 17, 2005, at 11:32 AM, Ruediger Pluem wrote: >
[..cut..] > > I still think that having this http "specific" error mode > "hidden" within the core output filter is misguided. Instead, > a specific http output filter is, imo, a better place. Sounds reasonable. I think the code can be placed in http_filters.c and the filter itself could be added by ap_http_header_filter once it finished its own work (just the way it currently does with the chunk filter). > > Even so, I say let's fold this into trunk as is, and > then work on abstracting it out. Yes, I will do so. Even more I will propose it for backport. Although this fix is not "perfect" with respect to your remarks above it fixes an actual problem without being a total nasty hack and thus should be included in the 2.2.x branch. If it we manage to extract this to a separate filter before 2.2.1 I would be even more happy. Regards Rüdiger
