Path: utzoo!attcan!uunet!husc6!uwvax!oddjob!ncar!ames!pasteur!ucbvax!bloom-beacon!ATHENA.MIT.EDU!swick
From: swick@ATHENA.MIT.EDU (Ralph R. Swick)
Newsgroups: comp.windows.x
Subject: Re: upper case fontnames
Message-ID: <8805171244.AA19486@LYRE.MIT.EDU>
Date: 17 May 88 12:44:50 GMT
References: <8805162130.AA14397@godzilla.ele.toronto.edu>
Sender: daemon@bloom-beacon.MIT.EDU
Organization: DEC/MIT Project Athena
Lines: 27


     Date:  Mon, 16 May 88 17:30:08 EDT
     From:  Mark Moraes <moraes%godzilla.ele.toronto.edu@RELAY.CS.NET>
     
     Just curious - why are all fontnames converted to lowercase by the 
     servers? (See server/os/{4.2bsd,hpux} for examples.

     The server documents mention in passing (where they describe
     ExpandFontName, I think) that the fontnames should be
     case-insensitive. Shouldn't that match both lower and upper case?

Implementation hack (or bug, depending on your frame of reference).
The Protocol requires fontnames to be case insensitive and contain
only ISO Latin-1 characters, but the implementation (and meaning of)
fontnames is not otherwise constrained.

The sample server chooses to store fonts in individual files and the
most efficient way to implement a case insensitive filesystem lookup
is to single-case all fontnames.

Remember that servers tell clients what fonts they support (via
XListFonts), and there is no requirement that this list be piece-wise
modifiable by a user.  An implementation that hardcoded fonts internally
or that stored multiple fonts in a single file or that computed them
dynamically could still be a conforming implementation.

-Ralph