[Lispweb] Mod_lisp 2.0 released

[Daniel Barlow]

Craig Brozefsky <craig at red-bean.com> writes:

> The model has some concrete advantages:
> 
[...]

As long as we're discussing the general approach:

8.  Dynamic but infrequently changing content can be cached by the
front-end webserver and future requests for the same page served
directly by it (using sendfile() or mmap() or something fast) without
having to involve Lisp - or involve Lisp only to ask "has this page
changed since {n}"

This is something that mod_lisp could score a significant win on if it
also added a cache control protocol so that Lisp could tell it to
discard outdated pages whenever it needed to.  You can do a certain
amount with mod_proxy (it claims, though I've never actually managed
to persuade it to do so in the real world) by setting appropriate HTTP
headers (Last-Modified, Expires, etc) and correctly handling
conditional GETs in your backend server.  But it would still be a lot
neater if you could control page flushing from the Lisp side instead
of the frontend having to poll.


-dan

-- 

  http://ww.telent.net/cliki/ - Link farm for free CL-on-Unix resources