always_direct confusion

Hi

I'm trying to get squid to not forward html requests to cache_peers, as I
suspect some of my peers (sibling, no-query, cache-digests only) have
refresh_patterns that are giving my cache stale versions of rapidly changing
(ie news) web sites.

That is, with gifs, jpgs etc I want squid to retrieve from peers if
available, but blah.html or http://www.blah.com/ I want to always obtain
from the source.

Is the below the correct way of doing this?

acl html urlpath_regex -i !\.gif$ !\.jpg$ !\.zip$ !\.class$ !\.jpeg$ !\.mid$
!\.exe$ !\.wav$ !\.cab$ !\.txt$ !\.mov$ !\.mp3$ !\.mpg$ !\.pdf$ !\.avi$
!\.txt$ !\.bmp$ !\.doc$ !/.gz$ !\.tar$ !\.tgz$ !\.jpe$ !\.mpeg$ !\.tif$
!\.art$ !\.arj$ !\.lha$ !\.mp2$ !\.qt$ !\.sit$
always_direct allow html

I'm finding the above still gets http://www.blah.com/ from a peer. Any idea
what I'm doing wrong?

Squid Cache: Version 2.3.STABLE3

Thanks

--
Regards
______________________________________________
Ian Chaloner
Systems Administrator
WestNet - WA's Statewide Internet Provider
Phone: 9218 2600 - Fax: 9218 2666
http://www.wn.com.au
______________________________________________