Re: Re: Content-length header

Joerg Sommer wrote:
> "Paul A. Rombouts" <[email protected]> wrote:
>>In the version of WWWOFFLE that I maintain for personal use, I try to 
>>retain the Content-Length header or even add one when the length of the 
>>response to the client can be known in advance. I do not use chunked 
>>encoding in this case.
> 
> 
> Would you publish this patch?

I publish my WWWOFFLE patches at 
http://www.phys.uu.nl/~rombouts/wwwoffle.html
and
http://members.home.nl/p.a.rombouts/wwwoffle.html

These are large patches and contain many modifications besides the one 
you are interested in. You should read the instructions in README.par 
carefully and decide whether you find all these customizations 
acceptable. Maintaining all these features as separate patches would be 
too much work for me, so it is an all-or-nothing deal.

> I'm not familiar with the wwwoffle code,
> but where is the problem? Doesn't wwwoffle has an outgoing queue or a
> send module, where the content-header become generated?

There are two cases where my version of WWWOFFLE sends a Content-Length
header to the client:

- WWWOFFLE is online and the remote server provides a Content-Length
   header and the content is not compressed/decompressed nor modified.
- WWWOFFLE serves a file from the cache and the content is not
   compressed/decompressed nor modified.

In all others cases WWWOFFLE would have to store the whole content in a 
temporary file first to provide a reliable Content-Length header, which 
would be inefficient.

-- 
Paul Rombouts