[Web-SIG] [ANN] Aspen 0.6 -- turning towards 1.0 w/ daemonization

[Chad Whitacre]

Matt,

Thanks for your questions!


Has there been any thought given to dealing with virtual hosting?


Yes. I hope to implement that as an Aspen application. Basically, my idea is
to put all virtually hosted websites in a directory, /usr/local/www say, and
start a main Aspen process in that dir which would maintain an Aspen child
process for each subdir/website. The main process would then proxy to its
children based on the Host header. Waddya think?


Reverse proxy?


With Aspen in which role, the proxy or the origin server? The above is
basically a reverse proxy (no?); Aspen could of course be proxied itself
behind Apache or Varnish or whatever.


Since this is WSGI based, can I deploy it under mod_python or say
> fcgi?  Or twisted?


Aspen is intended to be deployed in production all on its own. If you'd like
to deploy it otherwise, the place to start is the aspen.website.Websiteclass:

  http://aspen.googlecode.com/svn/tags/0.6/src/aspen/website.py

Unfortunately, the Python API will probably be the last thing documented,
sorry. :-/ Feel free to ask questions, probably on the aspen-users mailing
list:

  http://groups-beta.google.com/group/aspen-users/topics


Do you have any stats on how static performance is compared with say
> plain old apache?
>

There's a couple threads on this on the aspen-users list. The bottom line is
that Aspen is, um, slower than Apache at serving static content. :-)  You'll
notice, though, that we do have a patch for this already (thanks Giorgi!),
which may give us a 5x speed-up for static content. I haven't reviewed the
patch yet, and would be interested in any review or benchmarks you could
provide. :-)

In general, the 1.0 release is focused on the API. If Aspen succeeds, I
expect some parts will eventually be rewritten in C, but first things first.

Thanks again for your feedback!



chad
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.python.org/pipermail/web-sig/attachments/20061208/88651ddf/attachment.htm