httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From d...@ast.cam.ac.uk (David Robinson)
Subject Re: Harvest cache now available as an ``httpd accelerator'' (fwd)
Date Wed, 05 Apr 1995 17:18:00 GMT
> > Could we do the same thing in apache? Have a cache of URL -> file mappings?
>
> Now that sounds really useful, as long as there are no problems
> with mappings becoming out of date, but if it's part of apache, we
> should be able to control that.
> 
> I bet a big chunk of apache time is spent looking up the same
> URLs over and over and over...
> 
> Rob T and David have experience of writing code to cache client names,
> so do either of you fancy having a crack at it. Speak now to avoid
> duplicating effort.

I wouldn't mind giving it a go. But there are lots of questions; to start with:
What goes into the cache? The 'key' will presumably be the URL and content
negotiation data (precisely what?). What will the 'value' be? The actual data
last sent out for this request? (Maybe kept in a temp file.) Or, for
non-scripts, a reference to the original file?

Would httpd ever cache data without checking to see if the file had been
changed? (For timescales of minutes, this may be ok.)

If you do want to try and test for updates, what would you check?
The date and inode number of the file? The date of the mapfile, for content
negotiation? Changes to any .htaccess files? Clearly too much checking would
be no different to the normal httpd processing.

What about files subject to authentication?

Input please.

 David.

Mime
View raw message