commloops testing

hi guys,

I _think_ I found the niggling bug left in the expired object/IMS
magic in client_side.c, and after some polygraphing and local browser
user the commloops branch seems to be stable.

The commloops branch is basically as functional as the squid-HEAD branch,
minus range request caching (so a range request is always uncacheable
and is passed through to the server side vertabim atm).

Could I possibly get some quick third-party testing to happen?

I'd like to look at committing the current work set to squid-HEAD
before I move onto the next phase of the commloops/modio work, so any
third-party testing/feedback I can get from the squid-dev list members
will be very appreciated!

Just for those who aren't keeping up - I've reworked the storeClientCopy()
call to actually be a stream rather than a copy from random parts of
an object. Once this code is stable, squid will be much closer to supporting:

  * event IO (/dev/poll, kqueue, native win32 completion port stuff)
  * content modification (since its a stream rather than a "paged object" now,
    it ties into rbcollins' content module work from 2000/2001)
  * better storage manager (too long and hard to explain here :-)

So I'm kind of hanging out to get this stuff tested, stable and committed
before I move onto the next phase.

Thanks!

Adrian
Received on Fri Jan 11 2002 - 23:06:05 MST