Re: autoconf 2.64 fails when IFS is changed

[Eric Blake]

According to Ralf Wildenhues on 7/28/2009 12:39 PM:
> Hello Arkadiusz,
> 
>> old_IFS=$IFS
>> IFS=","
>> AC_DEFINE([VC_ENABLE_API_COMPAT],   [1])
>> IFS="$old_IFS"
>>
>> AC_OUTPUT
>> [ar...@carme-pld ~/test/1]$ ./configure
>> ./configure[1640]: print -r --: not found
>> configure: creating ./config.status
>>
> Thanks for the report.  This will probably mean that we will document
> that you should not change, or that you need to restore, IFS when
> invoking Autoconf macros.  In general, we really cannot guarantee which
> macros will expand to IFS-sensitive code and which won't, so you may
> have to assume they all will.

Cleaning out some old mails in my inbox.  I agree with Ralf that the best
we can do is document IFS better (it is easier to do that than to ask all
internal uses of $* to reset IFS to the defaults).  I will be pushing this.

-- 
Eric Blake   ebl...@redhat.com    +1-801-349-2682
Libvirt virtualization library http://libvirt.org

From 2c0a2e14f09a363d9fddd1d31147e84746069726 Mon Sep 17 00:00:00 2001
From: Eric Blake <ebl...@redhat.com>
Date: Thu, 25 Feb 2010 14:06:38 -0700
Subject: [PATCH] Document that Autoconf relies on IFS.

* doc/autoconf.texi (Special Shell Variables) <IFS>: Add details
about use of IFS within configure script.
* THANKS: Update.
Reported by Arkadiusz Miskiewicz.

Signed-off-by: Eric Blake <ebl...@redhat.com>
---
 ChangeLog         |    8 ++++++++
 doc/autoconf.texi |   10 ++++++++--
 2 files changed, 16 insertions(+), 2 deletions(-)

diff --git a/ChangeLog b/ChangeLog
index b7caa93..fa477a1 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,11 @@
+2010-02-25  Eric Blake  <ebl...@redhat.com>
+
+       Document that Autoconf relies on IFS.
+       * doc/autoconf.texi (Special Shell Variables) <IFS>: Add details
+       about use of IFS within configure script.
+       * THANKS: Update.
+       Reported by Arkadiusz Miskiewicz.
+
 2009-09-16  Eric Blake  <e...@byu.net>

        Optimize AC_REPLACE_FUNCS.
diff --git a/doc/autoconf.texi b/doc/autoconf.texi
index e5cf9fd..8f44ffe 100644
--- a/doc/autoconf.texi
+++ b/doc/autoconf.texi
@@ -15715,18 +15715,24 @@ Special Shell Variables
 but this caused many problems so modern shells ignore any environment
 settings for @env{IFS}.

-Don't set the first character of @code{IFS} to backslash.  Indeed,
+Don't set the first character of @env{IFS} to backslash.  Indeed,
 Bourne shells use the first character (backslash) when joining the
 components in @samp{"$@@"} and some shells then reinterpret (!)@: the
 backslash escapes, so you can end up with backspace and other strange
 characters.

-The proper value for @code{IFS} (in regular code, not when performing
+The proper value for @env{IFS} (in regular code, not when performing
 splits) is @sa...@key{spc}@key{t...@key{ret}}.  The first character is
 especially important, as it is used to join the arguments in @samp{$*};
 however, note that traditional shells, but also bash-2.04, fail to adhere
 to this and join with a space anyway.

+M4sh guarantees that @env{IFS} will have the default value at the
+beginning of a script, and many macros within autoconf rely on this
+setting.  It is okay to use blocks of shell code that temporarily change
+the value of @env{IFS} in order to split on another character, but
+remember to restore it before expanding further macros.
+
 @item LANG
 @itemx LC_ALL
 @itemx LC_COLLATE
-- 
1.6.6.1

signature.asc
Description: OpenPGP digital signature