Re: URL_REGEX case insensitive?

On Tue, 29 Apr 1997, Arjan de Vet wrote:

> In article
> <Pine.LNX.3.95.970429194932.20310B-100000@amarok.glasswings.com.au> you
> write:
>
> >What would be useful for this is possibly making a utility to scan news
> >messages for url's and removing those (using the util to scan newsgroups
> >like alt.binaries.pictures.erotica.whatever). However I think you'd run
> >into ACL lists that are so long squid slows down to a crawl. *IS* there a
>
> Use the splay tree or binary tree option in src/Makefile when using long
> ACL lists for blocking sites. And let me know if they work OK :-), Ed
> Knowles and I have put a lot of effort in it.
>
> >limit on the amount of ACL lists squid has?
>
> As far as I know all ACL's are dynamically created.

There is no hard-coded limit. However, I doubt very much that without a
lot of code changes you can put big lists into squid. In fact, we've
verified this with practical experience... :-).

Splay trees are just a fancy way of building binary trees. Since acls are
not dynamic once they have been loaded, splay trees are a waste of time.
In addition, simple binary trees are a very inefficient way of searching
URLs. They will certainly be better than a linear linked list, but not
fast enough for a large list on a production server. There are ways to do
very fast searches on large lists, but they require careful attention to
the problem space and result in extensive changes to the code.

Kevin

------------------------------------------------------------------------------
 Kevin Fink <kevin@fink.com> N2H2, Creators of Bess
 http://www.fink.com/ 1301 Fifth Avenue, Suite 1501
 http://www.n2h2.com/ Seattle, WA 98101
------------------------------------------------------------------------------
 (206) 971-1451 VOICE (206) 971-1460 FAX (206) 680-7666 PAGER
------------------------------------------------------------------------------
Received on Thu May 01 1997 - 20:44:40 MDT