Re: A couple URL problems...

Hi Clifton,

I know you are probably tired after surfing two months of list archives,
but here is one more thing for you to try:

> grep space squid.conf

HTH,

Alex.

On Wed, 16 Jun 1999, Clifton Royston wrote:

> Hi, all:
>
> I just looked over the last month or two's posts to the list, and
> didn't see anything about this subject.
>
> I'm running Squid 2.2STABLE3 in an alpha-test mode on our in-house
> network, using it as an explicitly configured proxy. (OS and hardware
> are probably irrelevant for now.) If this test goes well, I'm planning
> to do a brief test with transparent proxy on our in-house network,
> using a borrowed ServerIron, then install a dedicated Squid server - a
> suitably hefty BSD/OS 4.0 box, probably with RAID for uptime - do a
> formal customer beta-test, and finally start announcing the explicit
> proxy for customer use. Eventually, I would like us to go over to
> full-scale transparent caching via L4 switch redirection, but I feel we
> need to test this configuration fanatically first and make sure it is
> genuinely transparent in all cases.
>
> However... I have run into several sites now, mostly e-commerce
> sites, which are accessible directly via Netscape but not via Squid
> proxy. This is the kind of nagging "gotcha" that could become a
> serious problem in a transparent caching environment, where there's no
> way for the end-user to just turn it off. Here's one:
>
> <http://www2.tesys.com/telenetparts.store/284979100/Product/View/HD&20&20SSA18BLW/HD SSA18BLW>
>
> (Note the space in the last portion of the URL - blecch!!)
>
> Squid rejects this with:
>
> "Invalid URL
> Some aspect of the requested URL is incorrect. Possible problems:
> Missing or incorrect access protocol (should be `http://'' or similar)
> Missing hostname
> Illegal double-escape in the URL-Path
> Illegal character in hostname; underscores are not allowed "
>
> Is there a ready-made Squid patch to work around this?
>
> A similar problem occurs if you put in more than one search term in
> the store search pages at:
>
> <http://store.knifecenter.com/knifecenter/search.html>
>
> A different but related problem occurs at the same site with their
> keyword search if you put in a single term and get a multi-page list.
> In this case, the URL looks something like this:
>
> <http://store.knifecenter.com/pgi-KeywordSearch3?katana,Contains>
>
> and it presents a form with buttons; clicking the button sends one to
> the horrendous URL:
>
> <http://store.knifecenter.com/pgi-ContinueList1?Search For Items Containing ALL Of These Keywords:<BR>>
>
> (Yes, that's really what it apparently sends as the URL.)
>
> If I bypass the proxy, this actually works. I'm assuming that the
> spaces in the URL are not valid HTTP, though I haven't checked the
> RFCs, but it looks like I may need to work around this if multiple web
> sites are doing the same thing, broken or not.
>
> Comments or suggestions?
> -- Clifton
>
>
>
Received on Wed Jun 16 1999 - 23:24:49 MDT