On Mon, 2008-04-07 at 13:09 +0200, Arno _ wrote:

> I'm my configuration I have 2 bluecoat proxy talking to a webwasher via ICAP 
> to control the URL.
> And I also have a squid 3.0 for my test and some special production purpose.
> My squid is doing a limited authentication, using basic or none for some IP 
> range. I can't and don't want to do any other kind of  authentication as I do 
> on the production proxy.
>  
> So to be able to make it work with the ICAP server (webwasher in my case) I 
> need to send user name and user group to it so that I can control on the 
> Webwasher the URL accessed from the test user and some production server.
>  
> Since on the current squid (3 stable 4) there is nothing to let me cheat with 
> the ICAP entry I decide to add some feature to it.
> 
> I add the following ICAP option:
> icap_fake_client_username: let me specifies the client username that has to 
> be put into the icap-client-username ICAP header, applied only if the 
> icap_send_client_username is set; no default.
> 
> icap_client_group_header: let me create a header to be send into the ICAP 
> header, be default it's set to X-Client-Groups, and, for now, only used if 
> the next field is present
> 
> icap_fake_client_group: let me specifies the client group that has to be put 
> into the icap-client-group ICAP header; no default.
> 
> 
> TODO if possible: retrieve the client-group from the authentication 
> procedure, if done in NTLM, AD, LDAP or other method that will allow this 
> information. But I think it will need a lot of change
> 
> 
> The change are made to the following file:
> cf.data.pre
> ICAP/ICAPConfig.h
> ICAP/ICAPModXact.cc
> 
> 
> Sound interesting ?
> Someone willing to give me instruction on how to continue the job ?
> Suggestion on way to improve it ?

Please file a bug report with your patches. That way they will not be
forgotten and may be reviewed.

It would also be great if you could ask on squid-users whether anybody
ever felt the need to fake these ICAP headers. The patches are much more
likely to be accepted if it is a common need.

Thank you,

Alex.


Reply via email to