On Mon, Mar 31, 2014 at 09:50:14AM -0400, John David Anglin wrote:
> >It's also not a problem on x86, so this is a hppa-specific bug,
> >apparently:
> There is no documentation that I can find saying that dietlibc is
> ANSI/POSIX compatible and that it flushes and closes output streams
> on exit. There was some issue apparently with this in the past
> regarding ash.
>
> However, I see the parisc version of start.S branches to _exit,
> not exit. That's probably the reason for this behavior.
So I'm not unalterably opposed to adding a fflush(stdout) to two
places to help address this. But the issue is I'm sure this problem
is going to hit a *huge* number of places, and not just these.
I'd be happier making these two code paths call exit(0) explicitly,
which I think would solve the problem, but I'd strongly suggest filing
a bug against dietlibc, since it's going to cause other problems, and
from what you've said, and what I had guessed earlier, this is clearly
a hppa specific bug with dietlibc, and it's better fixed in dietlibc.
- Ted
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
