You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Greg Stein <gs...@lyra.org> on 2000/11/04 00:26:34 UTC
broken core_output_filter
I can't find the stupid CVS commit, but the recent change to
core_output_filter to look for FLUSH buckets broke it. It assumes that a
flush occurs at the end of a brigade. Unlike an EOS bucket (which has the
*semantics* of "end of content"), I don't believe that we can assume the
same for FLUSH buckets.
Ryan: I believe that was your change to the core_output_filter. I think the
right answer would be to set a flag that says "saw flush" rather than
breaking out of the loop.
Urk. Looking in there, I see we also assume a single FILE bucket per
brigade. That isn't going to work either...
Cheers,
-g
--
Greg Stein, http://www.lyra.org/
Re: broken core_output_filter
Posted by rb...@covalent.net.
On Sat, 4 Nov 2000, David Reid wrote:
> > Ryan: I believe that was your change to the core_output_filter. I think
> the
> > right answer would be to set a flag that says "saw flush" rather than
> > breaking out of the loop.
>
> I think it was Jeff's wasn't it? I know it's *normally* Ryan...
>
> :)
No, it was mine. :-)
Ryan
_______________________________________________________________________________
Ryan Bloom rbb@apache.org
406 29th St.
San Francisco, CA 94131
-------------------------------------------------------------------------------
Re: broken core_output_filter
Posted by David Reid <dr...@jetnet.co.uk>.
> Ryan: I believe that was your change to the core_output_filter. I think
the
> right answer would be to set a flag that says "saw flush" rather than
> breaking out of the loop.
I think it was Jeff's wasn't it? I know it's *normally* Ryan...
:)
david
Re: broken core_output_filter
Posted by rb...@covalent.net.
This filter is broken in a number of ways. I am currently working on
getting it to save a brigade past the end of a request, so that pipelining
works better. When I finish that I will re-visit the flush bucket
semantics. Depending on how adventurous I feel, I may just start this
filter over.
Ryan
On Fri, 3 Nov 2000, Greg Stein wrote:
> I can't find the stupid CVS commit, but the recent change to
> core_output_filter to look for FLUSH buckets broke it. It assumes that a
> flush occurs at the end of a brigade. Unlike an EOS bucket (which has the
> *semantics* of "end of content"), I don't believe that we can assume the
> same for FLUSH buckets.
>
> Ryan: I believe that was your change to the core_output_filter. I think the
> right answer would be to set a flag that says "saw flush" rather than
> breaking out of the loop.
>
> Urk. Looking in there, I see we also assume a single FILE bucket per
> brigade. That isn't going to work either...
>
> Cheers,
> -g
>
> --
> Greg Stein, http://www.lyra.org/
>
>
_______________________________________________________________________________
Ryan Bloom rbb@apache.org
406 29th St.
San Francisco, CA 94131
-------------------------------------------------------------------------------