-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Paul Eggert on 4/2/2007 12:41 PM:
>
> Anyway, for what it's worth I think fflush (stdin) should behave like
> 1988 POSIX and glibc, at least for apps that care about this sort of
> thing.
Here's my first attempt at this. I don't have ea
two lists of tools: one in [EMAIL PROTECTED] Utilities in Makefiles' and
one in
[EMAIL PROTECTED] Install Command Categories'. Should libtool now need to
restrict itself to the intersection of both lists, as it is invoked
both at build and install time
I believe "intersection" is
As an extreme example, we can't assume 'shuf' (recently added to
coreutils).
Of course.
'fold' and 'cut' are not nearly as extreme as that, but
they do tend to be less-used in portable software.
Is there any modern (or even not-so-modern) system that does not provide
fold and c
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Bruno Haible on 4/7/2007 4:23 AM:
> $ ./show-portability copysignf
> libcmacosx-10.3
> libcygwin cygwin
> libmcygwin
The show-portability script output is a bit confusing for cygwin. Yes,
the fuc
Hi Paul,
> Here's a proposed patch to 'signbit' to have it use 'copysign' if
> available.
Globally, I think this is backwards: 'copysign' is a more general function
than 'signbit', since it takes 2 arguments. And the usual way to return
a boolean result is in an integer register of the CPU, not i
