httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Greg Stein <gst...@lyra.org>
Subject Re: cvs commit: apache-2.0/src/main http_protocol.c
Date Thu, 09 Nov 2000 03:21:59 GMT
On Thu, Nov 09, 2000 at 03:00:02AM -0000, rbb@locus.apache.org wrote:
> rbb         00/11/08 19:00:02
> 
>   Modified:    src/main http_protocol.c
>   Log:
>   The content-length filter needs to compute the length even if the
>   content-length has already been set.  If we don't, then we won't know if
>   the content-length has changed because some other filter changed the
>   length of the actual data.

Euh... that just doesn't seem right. If a filter previous to this one knows
what the C-L is and sets its, then this filter should not monkey with that.

As a result of this change, then the C-L filter is going to buffer up the
whole response, even though somebody has already said "it is <this> long."

We need a better solution for content generators and filters to figure out
what the C-L is. If an upstream filter/generator determines the output
length and sets it, but then a downstream filter changes it, then should
that downstream filter remove the C-L header? What happens if the header has
been sent already?

But the change above, which is effectively, "I'm going to throw out what you
computed, and redo it myself" is wrong.

Cheers,
-g

-- 
Greg Stein, http://www.lyra.org/

Mime
View raw message