Re: Some important packages have been orphaned
On Fri, Jul 26, 2002 at 11:13:08AM +0200, Jérôme Marant wrote:
> > As for gnat 3.1, certain floating point operations are next to useless.
> > I could work around this by using the libc functions for sin, cos, etc,
> > but this seems a bit silly.
>
> This would be ugly. These are bugs and they have to be fixed.
>
> > (however, strangely enough, the actual source code is identical in both
> > versions).
>
> Do you mean the compiler code?
The source code which is actually i386 assembler (at least on my
machine).
See /usr/lib/gcc-lib/i386-linux/3.1.1/adainclude/a-numaux.adb.
However, I don't quite understand this code; if I take it and put it
on my program, it never works, not even on gnat 3.14. I suspect
that is my fault though.
> > (I am still a bit confused which version I should be trying to get
> > fixed).
>
> Well, gnat-3.1 has still not passed the ACATS tests. gnat 3.14p has.
> But, I think that everything should be done to improve gnat-3.1
> from now on. ACT said gnat-3.1 should be better than 3.14p.
Ok.
> > Actually I used the BTS system for reporting gcc bugs. See bugs 6910 and
> > 6911 or Debian Bugs 147145 and 148529 (respectively). The gcc BTS bugs
> > <URL:http://gcc.gnu.org/cgi-bin/gnatsweb.pl> are linked with the Debian
> > BTS bug entries.
>
> I've seen them all. I would expect ACT to speak up but what they plan
> is still quite unclear. 3.1 is work in progress, I hope 3.2 will
> be better.
>
> Did you try to compile with the CVS version from HEAD?
No, I haven't. I will have to try...
Can you run it from the source tree, or does it need to be installed?
> > I haven't reported the bug specific to gnat 3.14p upstream though.
> > Should I?
>
> I think you should, yes.
If development is now proceeding into 3.1, there may not be much point
now...
--
Brian May <bam@debian.org>
--
To UNSUBSCRIBE, email to debian-devel-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: