Re: Changing configuration without restarting

--MimeMultipartBoundary
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable

My current todo (stable for the last three weeks):

1. Fix the support for pipelined requests. The current code has a major
design flaw.

2. Generic support for requests of any type with any body size. The
current code is hardcoded on that GET requests never have a body, and
POST does.. Apart from some changes related to 1, this requires a some
nice way of streaming bodies, and a rewrite of POST/PUT handling to
support this.

3. When we have good generic support for requests with bodies, then
implement FTP PUT. When 2 is done this task is very easy (only a couple
of lines).

4 Better support for dynamic reconfiguration.

5 More structured method/protocol selection

6 Systematic testing of Squid. I am currently verifying persistent
connections but many other parts needs testing as well. There might well
be some other major design flaws like the ones found in the persistent
connection code. I'll probably rewrite my current testsuite developed to
test persistent connections into something useful later on. The current
testsuite consists of a simple tweaked web-server written as a
shell-script and some test data for the tests I am doing. I'll probably
remake this into perl and polish it up a bit.

Duane keeps his own Todo, not related to mine.

Priority order:

1 has very high priority.

2 is required for 3, and I'd like to do 3.

4 is wanted by many users and others, and I agree that this needs to be
done. Not a easy task, but I think some work has already been done by
some other people. Can't remember who it was thought.

6 has high priority, but I have to concentrate on one task at a time in
order to get anything done with the limited time I have to work on
Squid. This has to evolve as make different tests.

---
Henrik Nordstr=F6m
Dancer wrote:
> You've piqued my curiosity. What _are_ numbers one and two on your list=
? :)
--MimeMultipartBoundary--