Re: proposed patch for "Tests failed with LINENO." Autoconf bug

[Paul Eggert]

> From: Akim Demaille <address@hidden>
> Date: 30 Oct 2001 16:15:28 +0100
> 
> Thanks, it looks good to me.

OK, but I'm not sure what you mean by "it" (please see below).

> From: Akim Demaille <address@hidden>
> Date: 30 Oct 2001 16:20:47 +0100
> 
> Paul> True.  I'd forgotten about that.  Here's a proposed patch to fix
> Paul> this dead horse one more time.  It uses Sed exclusively, instead
> Paul> of Awk.  It also fixes some of the other bugs we've talked
> Paul> about, but not all of them.
> 
> This fix is wrong, what was proposed was better.

Sorry, I've lost context.  Which patch do you prefer and why?

I already installed the _AS_LINENO_WORKS patch
<http://mail.gnu.org/pipermail/autoconf-patches/2001-October/007977.html>
since you indicated that you liked it in
<http://mail.gnu.org/pipermail/autoconf-patches/2001-October/007979.html>.
This patch is needed even if we get rid of the Awk+Sed code entirely.

As you mentioned, the patch that Raja R Harinath proposed in
<http://mail.gnu.org/pipermail/autoconf-patches/2001-October/007981.html>
is inadequate, as it can exceed sed limitations.

But the latest patch that I proposed, namely
<http://mail.gnu.org/pipermail/autoconf-patches/2001-October/007982.html>
doesn't have this problem: it uses two small sed scripts.  Also, it
fixes some other LINENO bugs that we've already discussed
(e.g. $LINENOUGH, $LINENO $LINENO).  When you say "This fix is wrong"
it suggests that you don't like this latest patch, but I don't know
why that would be.


> If the only obstacle to releasing Autoconf is your fear wrt this, then
> please remove it.

I'm not worried about the Awk+Sed code anymore, as I don't think it
will ever be executed on any host of concern to the GNU project.
(Perhaps I shouldn't have tried to fix it, as that just seems to be
leading to confusion.  :-)

> It's becoming more complicated that it needs to.

No argument there....