Jonathan Larmour wrote:
> It seems reasonable. Although I don't see why you should change squid. Why
> not have a little helper app which makes the request to localhost:3128 in
> the correct format. It then returns the data back. Basically a squid proxy
> proxy :-). Obviously use a daemon based on select() rather than something
> from inetd.
>
> This could be distributed with squid, but doesn't have to be part of it, not
> least because its only relevant for the Linux implementation.
>
I must admit that your idea of a separate process is very nice, and
can help if the router is not the machine hosting squid. For example,
a small ISP could give to the customers an automatic access to a
neighbor cache.
This process could loop on an accept() and fork a child for each
request. The idea of a big select() and several active connections
frights me. Yes, I know, squid itself is a nice piece of software,
but I dont think I am able to cope with all the problems...
Thanks
Eric Dumazet
Received on Thu Sep 12 1996 - 09:55:54 MDT
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:32:59 MST