Re: What is ``access (NULL, whatever)'' supposed to do?

[Thomas Schwinge]

Hello!

On Tue, Apr 10, 2007 at 09:30:15AM -0700, Roland McGrath wrote:
> Sorry, I can't agree.  glibc on Linux also sometimes changes so that things
> that previously got EFAULT start crashing instead.

I can clearly see the point you're trying to make.  Quality of software,
including to not rely on unspecified behavior.  As one did with all that
`PATH_MAX' gunk.  (Actually, I have a proposal to make for `PATH_MAX' and
friends.  I will send it in a separate email.)

But then, Roland, please introduce the brigade of programmers to us that
you're keeping underhand and that are going to fix all those tiny
annoyances and work on getting (and that's sometimes even more tedious)
the upstream programmes to include such patches.

We (you, Neal, I, ...) have more important things to do than teaching the
git authors (Linus, ...) that ``access (NULL, ...)'' is unspecified.  At
least I have.  Speculating, what do you think will be the first thing to
happen if I send a patch to the git mailing list for patching that?
``Works for us, go away.''  After some discussion, they might eventually
accept it.

If I -- one day, but I doubt it -- don't find anything anymore to do,
then I'm willing to start dealing with such things.

I know all this may be a bit exaggerating, but that's the point I'm
trying to make (as well as Neal is, I suppose).


I just think that fixing _our problems_ is more important to us, than
fixing _other peoples's_ coding issues.  (While I'm -- of course! -- not
at all reluctant to help them, if they want it.)


Regards,
 Thomas

signature.asc
Description: Digital signature