Paul Vixie wrote:
>>>As far as I remember the FSF didn't accept the code (long time
>>>ago now) because they didn't like the way the gcc backend was
>>>decoupled from the rest of the compiler (avoiding the tainting
>>>of all the rest of the M3 sources with the license).
>>
>>Yes. See <http://gcc.gnu.org/ml/gcc/2000-06/msg00010.html>.
>
>
> there's no evidence in that thread that RMS ever ruled on the question,
> it's just suppositions as to how things probably would go. and tim mann
> is wrong on two critical points: there is forward-compatibility between
> M3's BSD-like license and the GPL, just not backward-compatibility.
what
> i mean by that is, there is nothing in the M3 license which would
prohibit
> a code fork where the forked code had the M3 license AND the GPL. so,
if
> FSF wanted M3 they could slap their own license on top of the M3 license
> and thus control the rights to their copy.
>
> i think it's time to try again, but i'd like to be the one to try, since
> i've fought license wars before and i seem to have a good grasp of the
> fine points. but before anyone tries, can we rename the file we want to
> donate to FSF, and rename the functionality? if all it's doing it
> offering a pipe to the backend, then it's language-independent and not
> part of the M3 system at all. can we call it "pipe_to_backend.c" or
> something like that? what i'd like to do is reduce the number of
> questions posed by the donation down to: does FSF want this
functionality
> in their compiler?
>
> i also note that in a post-EGCS world, there are more voices than RMS's
> who would weigh in on these questions.
gcc has grown, and so has the m3cg backend, which currently looks like
this (for gcc 3.4.5):
% wc m3-sys/m3cc/gcc-3.4.5/gcc/m3cg/*
1 6 512 m3-sys/m3cc/gcc-3.4.5/gcc/m3cg/CVS
111 435 3316 m3-sys/m3cc/gcc-3.4.5/gcc/m3cg/Make-lang.in
35 197 1246 m3-sys/m3cc/gcc-3.4.5/gcc/m3cg/config-lang.in
40 220 1524 m3-sys/m3cc/gcc-3.4.5/gcc/m3cg/lang-options.h
51 184 1406 m3-sys/m3cc/gcc-3.4.5/gcc/m3cg/lang-specs.h
71 235 1369 m3-sys/m3cc/gcc-3.4.5/gcc/m3cg/lang.opt
108 550 3924 m3-sys/m3cc/gcc-3.4.5/gcc/m3cg/m3-convert.c
323 1269 12190 m3-sys/m3cc/gcc-3.4.5/gcc/m3cg/m3cg.h
4599 14487 120344 m3-sys/m3cc/gcc-3.4.5/gcc/m3cg/parse.c
5339 17583 145831 total
We'll have to check if other changes besides the new m3cg sub-directory
were necessary, and of course we should try to donate the most recent
version (for gcc 4.1), which is not checked in yet. Antony Hosking,
who has done most of the recent gcc integration work, is currently
traveling, but I think you should talk to him directly when he's
back in a few days. There should be no problem with renaming any
offending file or procedure.
Olaf
|
