[7880] in www-talk@info.cern.ch
Re: ANNOUNCE: NCSA X Mosaic 2.6b1 Released
daemon@ATHENA.MIT.EDU (wmperry@spry.com)
Fri Apr 7 10:46:37 1995
Date: Fri, 7 Apr 1995 10:43:40 +0500
Errors-To: procmaster@www19.w3.org
Reply-To: wmperry@spry.com
From: wmperry@spry.com
To: Multiple recipients of list <www-talk@www10.w3.org>
Henrik Frystyk Nielsen writes:
>
> > > o HTTP header User-Agent now uses a slightly different format, for
> > > consistency with other platforms:
> > >
> > >User-Agent: NCSA Mosaic/2.6b1 (X11;SunOS 4.1.3 sun4m) libwww/2.12 modified
> > >
> >
> > Is it possible that this field will ever conform to the spec?
> > User-Agent: NCSA-X-Mosaic/2.6b1 SunOS/4.1.3 libwww/2.12-modified
>
> Well, it's not far from it. Everything within '(' and ')' is to be
> considered a comment which is not part of the field, but is allowed. So
> what is left is the two phrases:
>
> NCSA Mosaic/2.6b1 and libwww/2.12 modified
>
> Here the problem is that both contain a space which is not allowed in a
> token.
>
> A User-Agent field like
>
> User-Agent: NCSA-Mosaic/2.6b1 (X11; SunOS 4.1.3 sun4m) libwww/2.12-modified
>
> would be OK.
I proposed to the HTTP working group that this be changed to something
like:
User-Agent: NCSA Mosaic/2.6b1 ; library=libwww/2.12 modified ;
os=SunOS 4.1.3 sun4m ; windowsystem=X11R5 ;
other-fun=here
This follows in the tradition of MIME having information that modifies
the main value (text/html ; version=3.0) - the version of Mosaic is
considered to be the main value, and all the others are just slight
modifications. This also doesn't break old log engines that had to deal
with the severely broken way that NCSA mosaic used before.
A few people liked it, but seems to be an HTTP/1.1 issue.
-Bill P.
