RE: [jim@meyering.net: Re: Reported bad code in Emacs configure file]

RE: [jim@meyering.net: Re: Reported bad code in Emacs configure file]

[Daniel Ortmann]

The only difference was "--without-gcc".  (I am not setting the
emacs_cvs_speed_t variable; neither is anything in the system, according to
env | grep speed.)

Note the configure script location marked with @@@ below.  The "if" part
does set the speed variable; the "else" part does not.  And, since I am not
setting it myself, I believe this is why the variable is not being set.

In other words, I believe that, when using the "--without-gcc" option, the
"if" part is not being run and therefore the variable is not being set.

    echo $ac_n "checking for speed_t""... $ac_c" 1>&6
    echo "configure:3020: checking for speed_t" >&5
    if eval "test \"`echo '$''{'emacs_cv_speed_t'+set}'`\" = set"; then
      echo $ac_n "(cached) $ac_c" 1>&6
    else
      cat > conftest.$ac_ext <<EOF
    #line 3025 "configure"
    #include "confdefs.h"
    #include <termios.h>
    int main() {
    speed_t x = 1;
    ; return 0; }
    EOF
    if { (eval echo configure:3032: \"$ac_compile\") 1>&5; (eval
$ac_compile) 2>&5; }; then
      rm -rf conftest*
@@@   emacs_cv_speed_t=yes
    else
      echo "configure: failed program was:" >&5
      cat conftest.$ac_ext >&5
    fi
    rm -f conftest*
    fi

    echo "$ac_t""$emacs_cv_speed_t" 1>&6
    if test "$emacs_cv_speed_t" = yes; then
      cat >> confdefs.h <<\EOF
    #define HAVE_SPEED_T 1
    EOF

    fi


Is there anything else I might be able to check?

-----Original Message-----
From: Richard Stallman [mailto:rms@gnu.org]
Sent: Friday, May 03, 2002 1:26 PM
To: dortmann@lsil.com
Subject: [jim@meyering.net: Re: Reported bad code in Emacs configure
file]


Can you talk with them and together figure out what's really going on?

------- Start of forwarded message -------
To: rms@gnu.org
Cc: emacs-devel@gnu.org, Alexandre Oliva <oliva@lsd.ic.unicamp.br>,
   Akim Demaille <akim@epita.fr>, Tom Tromey <tromey@cygnus.com>,
   Paul Eggert <eggert@twinsun.com>, Pavel Roskin <proski@gnu.org>
Subject: Re: Reported bad code in Emacs configure file
In-Reply-To: <200205010714.g417Epo07431@aztec.santafe.edu> (Richard
Stallman's message of "Wed, 1 May 2002 01:14:51 -0600 (MDT)")
From: Jim Meyering <jim@meyering.net>
Date: Wed, 01 May 2002 09:26:18 +0200

That code comes from configure.in:

dnl Check for speed_t typedef.
AC_CACHE_CHECK(for speed_t, emacs_cv_speed_t,
AC_TRY_COMPILE([#include <termios.h>], [speed_t x = 1;],
  emacs_cv_speed_t=yes, emacs_cv_speed_t=no))
if test $emacs_cv_speed_t = yes; then
  AC_DEFINE(HAVE_SPEED_T, 1, [Define to 1 if `speed_t' is declared by
<termios.h>.])
fi

Since that variable is defined either to no or to yes just above
(in the absence of a cached value), I don't see a need for double
quotes, unless he's doing something strange:

  -- he is using a corrupted cache file, or
  -- he has set emacs_cv_speed_t to the empty string in his environment

...
> *** configure.~1~       Fri Mar 15 06:46:09 2002
> --- configure   Tue Apr 30 13:26:28 2002
> ***************
> *** 3039,3045 ****
>   fi
>
>   echo "$ac_t""$emacs_cv_speed_t" 1>&6
> ! if test $emacs_cv_speed_t = yes; then
>     cat >> confdefs.h <<\EOF
>   #define HAVE_SPEED_T 1
>   EOF
> --- 3039,3045 ----
>   fi
>
>   echo "$ac_t""$emacs_cv_speed_t" 1>&6
> ! if test "$emacs_cv_speed_t" = yes; then
>     cat >> confdefs.h <<\EOF
>   #define HAVE_SPEED_T 1
>   EOF
------- End of forwarded message -------

Re: [jim@meyering.net: Re: Reported bad code in Emacs configure file]

[Paul Eggert]

> From: "Daniel Ortmann" <dortmann@lsil.com>
> Date: Fri, 3 May 2002 13:43:06 -0500
> 
> In other words, I believe that, when using the "--without-gcc" option, the
> "if" part is not being run and therefore the variable is not being set.

That's a bug in the GNU Emacs 21.2 configure.in script.  The following
patch has already been applied to the trunk, so I think the point is
moot.  I don't know who did the patch or when they did it, but it was
probably after 21.2 came out.

--- configure.in.~1.267.4.8.~	2002-03-15 04:45:21.000000000 -0800
+++ configure.in	2002-05-03 13:31:53.453360000 -0700
@@ -1362,7 +1362,7 @@ AC_TYPE_SIGNAL
 dnl Check for speed_t typedef.
 AC_CACHE_CHECK(for speed_t, emacs_cv_speed_t,
 AC_TRY_COMPILE([#include <termios.h>], [speed_t x = 1;],
-  emacs_cv_speed_t=yes))
+  emacs_cv_speed_t=yes, emacs_cv_speed_t=no))
 if test $emacs_cv_speed_t = yes; then
   AC_DEFINE(HAVE_SPEED_T)
 fi

RE: [jim@meyering.net: Re: Reported bad code in Emacs configure file]

[Daniel Ortmann]

Darn this company firewall anyway!  :-(   I've tried *all* of the tricks,
including using different ports (http ports, etc), and I just can't get
through.

I'll have to start taking a laptop home to get the most current CVS code
into work.

Or would it be possible to have the savannah emacs project provide a nightly
tar of the cvs trunk?  (caveat emptor, of course.)  _Then_ I could get
through the firewall.

-----Original Message-----
From: Paul Eggert [mailto:eggert@twinsun.com]
Sent: Friday, May 03, 2002 3:40 PM
To: dortmann@lsil.com
Cc: emacs-devel@gnu.org; oliva@lsd.ic.unicamp.br; akim@epita.fr;
tromey@cygnus.com; proski@gnu.org; rms@gnu.org
Subject: Re: [jim@meyering.net: Re: Reported bad code in Emacs configure
file]


> From: "Daniel Ortmann" <dortmann@lsil.com>
> Date: Fri, 3 May 2002 13:43:06 -0500
>
> In other words, I believe that, when using the "--without-gcc" option, the
> "if" part is not being run and therefore the variable is not being set.

That's a bug in the GNU Emacs 21.2 configure.in script.  The following
patch has already been applied to the trunk, so I think the point is
moot.  I don't know who did the patch or when they did it, but it was
probably after 21.2 came out.

--- configure.in.~1.267.4.8.~	2002-03-15 04:45:21.000000000 -0800
+++ configure.in	2002-05-03 13:31:53.453360000 -0700
@@ -1362,7 +1362,7 @@ AC_TYPE_SIGNAL
 dnl Check for speed_t typedef.
 AC_CACHE_CHECK(for speed_t, emacs_cv_speed_t,
 AC_TRY_COMPILE([#include <termios.h>], [speed_t x = 1;],
-  emacs_cv_speed_t=yes))
+  emacs_cv_speed_t=yes, emacs_cv_speed_t=no))
 if test $emacs_cv_speed_t = yes; then
   AC_DEFINE(HAVE_SPEED_T)
 fi

Re: [jim@meyering.net: Re: Reported bad code in Emacs configure file]

[Miles Bader]

"Daniel Ortmann" <dortmann@lsil.com> writes:
> Darn this company firewall anyway!  :-(   I've tried *all* of the tricks,
> including using different ports (http ports, etc), and I just can't get
> through.

What are the failure modes?  Does it provide _any_ access for users
(e.g., web browsing from your desktop machine)?

-Miles
-- 
80% of success is just showing up.  --Woody Allen

Re: [jim@meyering.net: Re: Reported bad code in Emacs configure file]

[Richard Stallman]

Thanks for clearing up the matter.