Re: Suggestion

David Luyer wrote:

> As for why it's not implemented - I guess nobody likes the
> idea enough to implement it, or people see some problem which
> I can't see...

I think the main problem is time. I am spending most of my free time on
finding the bugs in the 1.2 persistent connections support (and reading
all e-mail)... and when the persistent connections work as expected I
have some other major tasks that needs to be done in 1.2... But having
the URL (and related info) as a part of the swap file is a good idea by
several reasons:

* No need to keep the entire URL in memory
* Error detection (wrong swapfile object)
* Possible to recover from a lost swaplog
* Probably other reasons as well...

But what the development needs right now is some people that seriously
and systematically tests Squid. Not that much as a general proxy server,
but how it behaves in strange situations.

* Clients giving various malformed/strange requests
* Servers responding in strange/inconsistent ways
* Hunt for memory leaks
* What happens on high / to high load (and what can be done to ease
this)

And this testing is by far best done in a controlled environment, and
not using real life clients (programs is far easier to repeat). All
these tests requires that you can stop the server at any moment to do
some diagnostics, and it is most preferred if you can repeat the same
identical test later on.

---
Henrik Nordström