[132] in Athena User Interface
Re: RFC: Prioritized AUI component list
daemon@ATHENA.MIT.EDU (Richard Tibbetts)
Tue Jun 6 19:46:41 2000
Message-Id: <200006062346.TAA28050@hikari-no-ken.mit.edu>
To: tb@MIT.EDU (Thomas Bushnell, BSG)
cc: Richard Tibbetts <tibbetts@MIT.EDU>, Maciej Stachowiak <mjs@eazel.com>,
BSG@MIT.EDU, Bill Cattey <wdc@MIT.EDU>,
"Christopher D. Beland" <beland@MIT.EDU>, Brad Thompson <yak@MIT.EDU>,
aui@MIT.EDU, tibbetts@MIT.EDU
In-reply-to: Your message of "06 Jun 2000 18:49:03 EDT."
<u1hpupuxw68.fsf@pusey.mit.edu>
Date: Tue, 06 Jun 2000 19:46:34 -0400
From: Richard Tibbetts <tibbetts@MIT.EDU>
On 6/6 you wrote:
> Richard Tibbetts <tibbetts@MIT.EDU> writes:
>
> > On 6/6 tb@MIT.EDU (Thomas Bushnell, BSG) wrote:
> > > (Incidentally, as for AFS,
> > > don't assume that the existence of /afs means that the system is
> > > running AFS and don't use avoid-AFS-lossage behavior on non-AFS files
> > > [shell utils got both of these wrong]).
> >
> > As a resident free software geek, what is your suggestion for how to
> > deal with AFS from GPL'd code? I would prefer not to include ARLA
> > libraries in the Athena release. Do you have a better suggestion?
>
> Hmm. I'm really not sure. I think it's reasonable to interpret AFS
> as part of the operating system and detect its presence with autoconf
> and only use the functions if they are there. I don't know whether
> that actually passes muster with the language of the GPL. We should
> ask rms what he thinks, perhaps.
>
> I certainly hadn't thought about the problem from a licensing
> standpoint; that does raise a noxious wrinkle. Thanks for pointing it
> out.
My previous solution (which I implemented as a patch to gmc, though it
is not ready for prime time and likely won't get any more work, unless
AUI goes with gmc for some reason) was to shell out to afs commands.
This sucked, partly because their output and return values vary from
platform to platform.
A better (though far from perfect) solution would be to write a
non-free wrapper around the afs functions with good semantics and
parsable output, and then to shell out to that. How sketchy is that
with the GPL?
A third solutions (which I don't personally have much of a problem
with) is to disrespect the GPL, or try to squeak it by as "part of the
OS" as you say above.
tibbetts
-*- http://www.mit.edu/~tibbetts -*- finger tibbetts@monk.mit.edu -*-
