This is part of a fix to a recently-reported grep bug.
I've tested this with tar and coreutils too -- it doesn't
affect them, since they never use EXCLUDE_INCLUDE.
---
ChangeLog | 18
lib/exclude.c | 111 +++-
tests/te
Another I-hope-obvious patch.
---
ChangeLog |6 ++
lib/exclude.c | 17 +++--
2 files changed, 9 insertions(+), 14 deletions(-)
diff --git a/ChangeLog b/ChangeLog
index baff376..ae5ca9e 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,11 @@
2012-04-29 Paul Eggert
On 04/29/2012 03:31 PM, Bruno Haible wrote:
What kind of service from Autoconf, do you imagine, could help or simplify
this?
I was thinking of the case where a package uses Autoconf but not gnulib.
(Sorry, I don't have a solution for this problem, but I expect it is
a problem.)
Hi Paul,
> > The reason why we do this with explicit #ifs that test for a compiler
> > brand, not in an "Autoconfish way", is that we use the result in .h
> > files that should be installable
>
> That makes sense for the definition of _Noreturn in
> build-aux/snippet/_Noreturn.h, since that file
On 04/29/2012 03:22 PM, Bruno Haible wrote:
It is obvious how to extend tests/test-exclude4.sh or tests/test-exclude9.sh
to check against the problem?
It'd require more than that, since test-exclude.c would also
have to be changed. Please feel free ...
Paul Eggert asks:
> How about the following improvement to the heuristic?
Looks good to me.
Bruno
Hi Paul,
> I pushed this hopefully-obvious fix for problems I found
> while looking into a bug with grep and --exclude.
It is obvious how to extend tests/test-exclude4.sh or tests/test-exclude9.sh
to check against the problem?
Bruno
I pushed this hopefully-obvious fix for problems I found
while looking into a bug with grep and --exclude.
---
ChangeLog |8
lib/exclude.c | 33 ++---
2 files changed, 26 insertions(+), 15 deletions(-)
diff --git a/ChangeLog b/ChangeLog
index b81a4
Hi Dmitriy,
__author__ = \
[
'Bruno Haible',
'Simon Josefsson',
#'Dmitriy Selyutin',
]
Oops, I realize that Paul Eggert, who designed and coded the first
implementation of the --import functionality together with Simon, is
not listed. Fixed as follows in the old one:
2012-04-29 Bruno Hai
On Sun, 29 Apr 2012, Santiago Vila wrote:
> What could be the reason?
Oops! I forgot to run autoconf to update the configure script.
Now it works. Sorry for the noise.
On 04/29/2012 07:11 AM, Bruno Haible wrote:
The reason why we do this with explicit #ifs that test for a compiler
brand, not in an "Autoconfish way", is that we use the result in .h
files that should be installable
That makes sense for the definition of _Noreturn in
build-aux/snippet/_Noreturn.
On 04/29/2012 08:34 AM, Vincent Lefevre wrote:
I don't like the fact that you assume by default that compilers
are non-conforming
Nor do I. How about the following improvement to the heuristic?
It is just a heuristic so we can't do a perfect job, but the
following should be better than what we
Vincent Lefevre wrote:
> > we use the result in .h
> > files that should be installable and, after installation, used by a
> > potentially different compiler. Simon's and my packages use this feature.
> > The best way to get this done correctly are cascades of explicit #ifs.
>
> But there is a ris
On Sun, 29 Apr 2012, Bruno Haible wrote:
> This is not useful, because on Linux/PowerPC *is* broken.
> It would be better for you to incorporate the fix. Either by invoking
> "gnulib-tool --update", or by taking just the bug's workaround from
> the gnulib repository:
> http://git.savannah.gnu.org
On 2012-04-29 16:19:35 +0200, Bruno Haible wrote:
> Vincent Lefevre wrote:
> > modules/snippet/_Noreturn references build-aux/snippet/_Noreturn.h,
> > which is:
> >
> > #ifndef _Noreturn
> > # if (3 <= __GNUC__ || (__GNUC__ == 2 && 8 <= __GNUC_MINOR__) \
> > || 0x5110 <= __SUNPRO_C)
> > # d
On Sun, 29 Apr 2012, Bruno Haible wrote:
> This failure location indicates a severe problem with
>
> > In the meantime, I will disable this test in the Debian package.
>
> This is not useful, because on Linux/PowerPC *is* broken.
> It would be better for you to incorporate the fix. Either by i
On 2012-04-29 16:11:38 +0200, Bruno Haible wrote:
> Paul Eggert wrote:
> > it might make sense to try to do it in a more Autoconfish way.
>
> The reason why we do this with explicit #ifs that test for a compiler
> brand, not in an "Autoconfish way", is that we use the result in .h
> files that sho
Vincent Lefevre wrote:
> modules/snippet/_Noreturn references build-aux/snippet/_Noreturn.h,
> which is:
>
> #ifndef _Noreturn
> # if (3 <= __GNUC__ || (__GNUC__ == 2 && 8 <= __GNUC_MINOR__) \
> || 0x5110 <= __SUNPRO_C)
> # define _Noreturn __attribute__ ((__noreturn__))
> # elif 1200 <= _M
Paul Eggert wrote:
> it might make sense to try to do it in a more Autoconfish way.
The reason why we do this with explicit #ifs that test for a compiler
brand, not in an "Autoconfish way", is that we use the result in .h
files that should be installable and, after installation, used by a
potentia
Hello Santiago,
> I implicitly enabled gnulib tests in the Debian diffutils package
> (by switching debian/rules to debhelper) and now the package fails to
> build from source on the powerpc architecture because test-float fails:
>
> ./test-float
> test-float.c:333: assertion failed
> Aborted
>
Hello.
I implicitly enabled gnulib tests in the Debian diffutils package
(by switching debian/rules to debhelper) and now the package fails to
build from source on the powerpc architecture because test-float fails:
./test-float
test-float.c:333: assertion failed
Aborted
This is the relevant cod
21 matches
Mail list logo
