httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Greg Ames <grega...@raleigh.ibm.com>
Subject Re: woah, "GET /" with autoindex
Date Thu, 11 Jan 2001 21:48:04 GMT
Greg Stein wrote:
> 
> This kind of solution won't work. If you end up mixing ap_r* with bucket
> calls, then you're screwed. The bucket would hit the chain before the
> contents of the coalesce buffer.
> 
> And note that it would be quite easy for the mixture to happen. Handler H
> starts writing some content with ap_r*(). Then, it calls ap_cool_utility(r)
> which dumps some content via buckets. Oops!
>
good call...thanks!
 
> The answer is to optimize ap_r* to not require so much allocation. It might
> also be a good idea to insert a COALESCE filter IFF ap_r* gets called (e.g.
> insert the filter on the first ap_r* call).
>

Sorry, but I don't believe that's the _only_ answer, or optimal either. 
Perhaps something like an ap_rflush triggered somehow by this mixture of
calls you mentioned above.
 
> There was a question about whether other modules use ap_r* for lots of bits.
> Absolutely... mod_dav is a simple example. PHP uses ap_r*, too.
>
 ooooops...my apologies for not being more clear.  I was trying to ask
the opposite question: would there be any known cases of lots of
"tinygrams" going thru buckets directly, once we do the Right Thing in
ap_r* ?

...but the mod_dav case is interesting also.  Wouldn't that primarily be
big data (file uploads/downloads) mixed with a few small handshake type
packets?

Thanks,
Greg

Mime
View raw message