Re: Faster AT_CHECK, one less XFAIL

[Ralf Wildenhues]

Hi Eric,

* Eric Blake wrote on Fri, Nov 21, 2008 at 06:11:07AM CET:
> According to Eric Blake on 11/20/2008 7:19 PM:
> > Now that m4_expand is more robust, I switched AT_CHECK to start using it.
> >  I had to fix a bit of fallout - a grand total of two tests (one with
> > unbalanced '(', and one that intentionally mucked with the internals of
> > m4_split/m4_expand); not a bad track record.  It gives a lot more
> > assurance that m4_expand is ready for prime-time usage.  And it fixes a
> > LONG-standing XFAIL!  I'm pushing this now, so it can give m4_expand even
> > more stress testing.
> 
> And as long as I'm playing with autotest, this shaves off half a meg from
> autoconf's testsuite (17%).  It also fixes an echo usage bug if you name
> one of your files 'tests/-name.at'; conveniently hidden by the fact that
> it was 'echo AT_LINE' with no -, \, $ or ` until after you expand the m4
> macro, and thus easy to miss when grepping for problematic echo.

Nice fixes and size improvement!

One typo nit:

> --- a/lib/autotest/general.m4
> +++ b/lib/autotest/general.m4
> @@ -261,16 +261,43 @@ at_fn_banner ()

> +AS_FUNCTION_DESCRIBE([at_fn_check_prepare_dynamic], [COMMAND LINE],
> +[Decide if COMMAND at LINE is traceable at runtime, and call the
> +appropriate prepation function.])

preparation

Cheers,
Ralf