[EMAIL PROTECTED] (Eric Blake) writes:
> Do we want a consistent when doing a regex for both characters,
> where order doesn't matter; which is different than doing a literal
> like IFS where order does matter?
Yes we do. We want tab-space. That's the point of the maintainer rule.
I checked in the obvious fix.
2006-04-05 Paul Eggert <[EMAIL PROTECTED]>
* lib/autoconf/status.m4: Replace <space>''<tab> with
<tab><space> where this makes sense.
--- lib/autoconf/status.m4 5 Apr 2006 20:55:14 -0000 1.91
+++ lib/autoconf/status.m4 6 Apr 2006 02:28:56 -0000 1.92
@@ -574,7 +574,7 @@ m4_define([_AC_OUTPUT_HEADER],
#
dnl Quote, for the `[ ]' and `define'.
[ ac_dA='s,^\([ ]*#[ ]*\)[^ ]*\([ ][ ]*'
- ac_dB='\)[ '' (].*$,\1define\2'
+ ac_dB='\)[ (].*$,\1define\2'
ac_dC=' '
ac_dD=' ,']
dnl ac_dD used to contain `;t' at the end, but that was both slow and
incorrect.
@@ -606,14 +606,14 @@ uniq confdefs.h |
sed -n '
t rset
:rset
- s/^[ '' ]*#[ '' ]*define[ '' ][ ]*//
+ s/^[ ]*#[ ]*define[ ][ ]*//
t ok
d
:ok
s/[\\&,]/\\&/g
s/[\\$`]/\\&/g
- s/^\('"$ac_word_re"'\)\(([^()]*)\)[ ''
]*\(.*\)/${ac_dA}\1$ac_dB\2${ac_dC}\3$ac_dD/p
- s/^\('"$ac_word_re"'\)[ ''
]*\(.*\)/${ac_dA}\1$ac_dB${ac_dC}\2$ac_dD/p
+ s/^\('"$ac_word_re"'\)\(([^()]*)\)[
]*\(.*\)/${ac_dA}\1$ac_dB\2${ac_dC}\3$ac_dD/p
+ s/^\('"$ac_word_re"'\)[ ]*\(.*\)/${ac_dA}\1$ac_dB${ac_dC}\2$ac_dD/p
' >>conftest.defines
]
# Remove the space that was appended to ease matching.
