W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > January to April 1997

Re: ID: Proxy autoconfig

From: David W. Morris <dwm@xpasc.com>
Date: Thu, 27 Mar 1997 12:35:00 -0800 (PST)
To: "nemo/Joel N. Weber II" <devnull@gnu.ai.mit.edu>
Cc: dmk@bell-labs.com, josh@netscape.com, http-wg@cuckoo.hpl.hp.com
Message-Id: <Pine.SOL.3.95.970327122749.10539B-100000@shell1.aimnet.com>


On Thu, 27 Mar 1997, nemo/Joel N. Weber II wrote:

>    Date: Thu, 27 Mar 1997 09:38:06 -0500
>    From: Dave Kristol <dmk@bell-labs.com>
> 
>    A question about auto-config:  under what circumstances does it come into play?  For
>    example, suppose I start up a browser and it configures itself.  Then I fuss with
>    the configuration, use the browser for awhile, then quit.  When I start it again, is
>    the browser going to override the settings I put in place, or does it try to
>    auto-config only if there are no other settings?
> 
>    If the browser only tries to configure itself if there are no pre-settings, there's
>    another problem.  Suppose, given the scenario above, that I decide, for whatever
>    reason, that I want no proxies.  When I exit and restart, will auto-config force a
>    proxy on me again?
> 
> I think the answer should be that by default, a browser will try automatic
> configuration, and if that fails will use no proxy.
> 
> However, the preferences of the program should allow you to tell it
> to use something other than the default setting.  So you can either
> choose a proxy manually or disable proxies.

I think the default should be up to the client publisher. I would
recommend that once the user has manually altered setting, either those
setting persist or the user should be asked if the settings are temporary.

In the context of Netscape 3.0, the user has three proxy choices:
   1.  No
   2.  Manual ... which allows mix and match
   3.  PAC URL

The shipping default is NO proxy. I would expect that the new protocol
would either replace #3 or add a #4 and I would guess that the install
script would include an autoconfig option and either select #1 or #4 but
I mention these only to put in context the ID as I read it with how it
might be used. I don't see any need for the RFC which might result from
this dealing with how individual implementations might be seen by the
user except as setting the context for understanding the protocol.

Dave Morris
Received on Thursday, 27 March 1997 12:43:35 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:32:33 EDT