RE: Check for ar (libtool bug)

Ok, I'll buy that, but it just reminds me that I need to
go through and refactor libtool to make these things consistent. ;0
Approved and checking in to CVS head.

Robert

-----Original Message-----
From: Schleicher Ralph (LLI) [mailto:ralph.schleicher@xxxxxxxxxxxxxxxx]
Sent: Monday, February 24, 2003 8:29 AM
To: Robert Boehne
Cc: libtool-patches@xxxxxxx; libtool@xxxxxxx
Subject: Re: Check for ar (libtool bug)


Robert Boehne wrote:

> Typically we define things to ":" when they are not found.

Hi Robert,

$ grep AC_CHECK_TOOL libtool.m4
AC_CHECK_TOOL(RANLIB, ranlib, :)
AC_CHECK_TOOL(STRIP, strip, :)
  AC_CHECK_TOOL(DLLTOOL, dlltool, false)
  AC_CHECK_TOOL(AS, as, false)
  AC_CHECK_TOOL(OBJDUMP, objdump, false)
[AC_CHECK_TOOL(GCJ, gcj, no)
[AC_CHECK_TOOL(RC, windres, no)

That makes two times ":" (true) and 5 times "false"
or "no".

> If we change your patch to AC_CHECK_TOOL(AR, ar, :), does
> it still work to fix the problem you ran into?

It still works if "ar" is found but using ":" as the default
command for creating an old archive is IMHO the wrong thing.
Using ":" for "ranlib" and "strip" is acceptable because
libtool works well without them but AR=: returns the wrong
exit status if you attempt to create an old archive.
Therefore AR=false is correct.

-- 
Ralph



_______________________________________________
Libtool-patches mailing list
Libtool-patches@xxxxxxx
http://mail.gnu.org/mailman/listinfo/libtool-patches