[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Various problems compiling packages (+ binutils broken?)

Paul Slootman <paul@wau.mis.ah.nl> writes:

> On Tue 27 Oct 1998, Brederlow wrote:
> > I comiled (or tried to) several packages for Alpha and hit some
> > problems here and there:
> > 
> > X11: lots of "conversion for pointer to int of different size"
> > also in many other packages
> > 
> > Would you say that those warnings are bugs of the package, because it
> > makes them possible crash on "int != void *" archs?
> I've seen that many X toolkits misuse pointers to sometimes pass
> e.g. boolean values. On these, the warnings aren't a real problem
> (besides the fact this usage isn't efficient; they should use
> unions instead).

I noticed that too, but its still always a potential cause of a
bug. Its also bad programming. I think I will report bugs for every
such conversion.

> > config.guess: alphapca56-unknown-linux-gnu
> > 
> > Several programs using config.guess can't handle alpha flavours and
> > most (if not all) don't care anyway. Maybe config.guess should just
> > say alpha-unknown-linux-gnu there.
> Try the new autoconf package, things should be better in that respect
> (config.guess generated alphapca56-unknown-linux-gnu which wasn't
> understood by the config.sub or whatever; that should now be fixed).

OK, I will do that.

> > fakeroot: unaligned traps
> >           segfaults (specially with "fakeroot debian/rules binary")
> >              That might be a "conversion of pointer to int" problem
> >              with libtrix.
> Hmm, I've seen plenty of unaligned traps from fakeroot (every time, in
> fact), but no segfaults. Maybe I'm using an older fakeroot.

I tried to debug that a bit and gdb tells me that it crashes in
"rtld.c", which is libc. I swa that some changes were made to that
file for the current libc, so I will try that one (as soon as a new
kernel is compiled and running).

May the Source be with you.

Reply to: