W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > January to April 1995

Re: Browser Developers (A Plea From BrowserWatch)

From: <wmperry@spry.com>
Date: Fri, 24 Mar 95 11:23 PST
Message-Id: <m0rsEws-00000KC@monolith>
To: "M. Hedlund" <march@europa.com>
Cc: Dave Garaffa <d-garaffa@ski.mskcc.org>, http-wg%cuckoo.hpl.hp.com@hplb.hpl.hp.com
M. Hedlund writes:
> 
> On Fri, 24 Mar 1995 wmperry@spry.com wrote:
> >   Shouldn't the User-Agent field follow 'normal' specs for headers and do
> > something like:
> > 
> > User-Agent: agent=NCSA Mosaic/2.0.0b1 ; platform=Macintosh/680x0
> >             library=WWWLib/3.0.x ; other=Whatever
> > 
> >   This would make it a little easier for people like Dave to parse the
> > string, without having to worry about order of the arguments, etc.  Plus
> > this would allow whitespace (but not ;) in the specifier.
> 
> Yes, I like Bill's proposal and I've suggested platform in the user-agent 
> field before.  This would, however, completely break current systems 
> (which tend to look for the text before the first / as user-agent name, 
> and can survive miscreant whitespace) that currently work with HTTP/1.0 
> requests.  Can we defer this to 1.1?

  Perhaps we could combine the two?

User-Agent: NCSA Mosaic/2.0.0b1 ; platform=Macintosh/680x0 ;
            library=WWWLib/3.0.x ; other=Whatever

  This would still let old services that look for whatever is before the
first / to still get the same results.  And actually is more MIME-y than my
first one, since the platform=xxxx, can now be considered
parameters/extensions to the base 'NCSA Mosaic' value.

  -Bill P.
Received on Friday, 24 March 1995 12:36:30 EST

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 06:31:14 EDT