RE: [squid-users] dansguardian as a patch for Squid

Robert,

        iCAP seems to gain popularity these days; Squid is likely to
end up supporting it, eventually. Would it be possible to use iCAP for
both external modules and internal, compiled-in modules? I think that
iCAP is transport-layer independent so we could just pass the request
information to internal modules using function calls as opposed to
sockets that have to be used for external stuff.

        If the above is feasible, one could write code that can be
used both as a stand alone program and as a compiled-in module, with
95% of the code and logic being exactly the same. Moreover, we could
provide a simple ICAP library that completely masks the distinction
(i.e., transport layer becomes the responsibility of a library).

        Just a thought...

Alex.

On Mon, 25 Jun 2001, Robert Collins wrote:

> This shouldn't wait - getting agreement/consensus on the
> architecture for this in squid is quite important. Feedback from
> folk like you as to how it works, what you need, what you'd like
> etc is also very important. A couple of hours from you looking at
> this could save a lot of time alter.
>
> Well i-CAP isn't linux specific. I do most of my squid hacking on
> openBSD|MS Windows :].
>
> http://www.i-cap.org/
> http://www.checkpoint.com/cvpopenspec/CVPOpenSpecification.pdf
>
> The archives have a description of what I've done, separate from what
> i-CAP is. i-CAP defines a protocol for accessing things like content
> filters/virus scanners from a proxy. I've written a set of hooks for
> squid, and a hook API that allow direct access to what squid is doing
> for in-process content processors - so my doco is the relvant bit :].
> (And it's in the archives - but I couldn't dig it up on a quick search
> :-/). So I've included a synopsis of what it does/can do at the end of
> this email.
Received on Mon Jun 25 2001 - 07:53:20 MDT