RE: AC_PREFIX_PATH, AS_BASENAME fixes and a new test!

[Paul Wagland]

Well, I sure am glad that my changes produced such a flurry of responses!
:-)

Anyway, yes you are right, this should of also been in my patch as well. I
only added the BASENAME_SED so that it had the same interfaces as DIRNAME*.

I also appreciate the changes you made to the sed to make it portable as
well!

One other thing, given the concerns with the [\\/] path issue, is it
worthwhile adding in a new set of checks to ensure that
(BASE|DIR)NAME(_(EXPR|SED))? (Note 1) work with both \\ and /, and possible
a mixture of the two?

Thanks for all the co-operation,
Paul

Note 1: Yes I know this is not a portable definition of the macros, but
rather a GNU extended one :-)

> -----Original Message-----
> From: Paul Eggert [mailto:address@hidden
> Sent: Tuesday, September 18, 2001 1:00
> To: address@hidden
> Cc: address@hidden
> Subject: Re: AC_PREFIX_PATH, AS_BASENAME fixes and a new test!
>
>
> The AS_BASENAME fix doesn't seem complete to me.
> AS_BASENAME should look like AS_DIRNAME, e.g. something like this:
>
>       # AS_BASENAME(PATHNAME)
>       # --------------------
>       # Simulate running `basename(1)' on PATHNAME, not all
> systems have it.
>       # Also see the comments for AS_DIRNAME.
>       m4_defun([AS_BASENAME_EXPR],
>       [m4_require([_AS_EXPR_PREPARE])dnl
>       $as_expr X[]$1 : '.*/\([^/][^/]*\)/*$' \| \
>                X[]$1 : '\([^/][^/]*\)/*$' \|
>                X[]$1 : 'X\(//\)$' \| \
>                X[]$1 : 'X\(/\)/*$' \| \
>                .     : '\(.\)'])
>
>       m4_defun([AS_BASENAME_SED],
>       [echo X[]$1 |
>           sed ['???']])
>
>       m4_defun([AS_BASENAME],
>       [AS_BASENAME_EXPR([$1]) 2>/dev/null ||
>       AS_BASENAME_SED([$1])])
>
> except you need to fill in the '???'.
>
> AS_BASENAME_SED should be to AS_BASENAME_EXPR
> as AS_DIRNAME_SED is to AS_DIRNAME_EXPR.
>
>