Eric Blake wrote:
> > Therefore my proposal is to
> > - create a new module 'unused-parameter', that inserts a definition of
> > _GL_UNUSED_PARAMETER_ in a header file, like 'link-warning' and 'arg-
> nonnull'
> > do,
> > - update unistr.h accordingly.
>
> Sounds reasonable. But I'd r
On 24/12/09 23:21, Eric Blake wrote:
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Bruno Haible on 12/24/2009 1:08 PM:
Either way we create macros.h and put ASSERT there
Done this this big commit, now.
Hmm. macros.h defines SIZEOF, and test-argv-iter.c defines
ARRAY_CARDINALIT
Hi,
Just got this weird warning while creating a testdir:
$ ./gnulib-tool --create-testdir --dir=/tmp/testdir --with-tests
unistr/u32-strcoll
...
configure.ac:164: warning: AC_COMPILE_IFELSE was called before
AC_USE_SYSTEM_EXTENSIONS
glm4/fcntl_h.m4:20: gl_FCNTL_O_FLAGS is expanded from...
glm4
Eric Blake wrote:
> macros.h defines SIZEOF, and test-argv-iter.c defines
> ARRAY_CARDINALITY, both of which are used to determine the size of an
> array. I kind of prefer the name in test-argv-iter (to me, sizeof implies
> a size_t count of bytes, not the number of elements). Is it okay to
> ren
Eric Blake writes:
> Hmm. macros.h defines SIZEOF, and test-argv-iter.c defines
> ARRAY_CARDINALITY, both of which are used to determine the size of an
> array. I kind of prefer the name in test-argv-iter (to me, sizeof implies
> a size_t count of bytes, not the number of elements). Is it okay
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Eric Blake on 12/24/2009 4:21 PM:
> According to Bruno Haible on 12/24/2009 1:08 PM:
Either way we create macros.h and put ASSERT there
>> Done this this big commit, now.
>
> Hmm. macros.h defines SIZEOF, and test-argv-iter.c defines
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Bruno Haible on 12/24/2009 1:08 PM:
>>> Either way we create macros.h and put ASSERT there
>
> Done this this big commit, now.
Hmm. macros.h defines SIZEOF, and test-argv-iter.c defines
ARRAY_CARDINALITY, both of which are used to deter
Eric Blake wrote:
> > For me, with -Wall, plain 0 as a statement provokes a warning
> > "statement with no effect", but writing (void) 0 suppresses that
> > warning.
>
> Which implies the need for:
>
> # define SET_BINARY(fd) ((void) 0)
Yes, I agree. And when the result type is 'void', it should
Eric Blake wrote:
> Any objections to a patch that
> makes replacement of these headers unconditional, so that
> GNULIB_POSIXCHECK will work better?
>
> modules/arpa_inet:BUILT_SOURCES += $(ARPA_INET_H)
> modules/ctype:BUILT_SOURCES += $(CTYPE_H)
> modules/dirent:BUILT_SOURCES += $(DIRENT_H)
> mod
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Bruno Haible on 12/24/2009 12:40 PM:
> On Linux/glibc systems, I get this warning:
>
> test-pty.c:24: warning: initialization from incompatible pointer type
> test-pty.c:26: warning: initialization from incompatible pointer type
>
> It's
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Ben Pfaff on 12/24/2009 12:00 PM:
> Jim Meyering writes:
>
>> -# define SET_BINARY(fd) /* nothing */
>> +# define SET_BINARY(fd) 0
>
> For me, with -Wall, plain 0 as a statement provokes a warning
> "statement with no effect", but writi
Now that gnulib's and don't include any more,
I'm seeing this warning:
openat-proc.c: In function 'openat_proc_name':
openat-proc.c:91: warning: implicit declaration of function 'close'
Jim, is it ok to apply this?
2009-12-24 Bruno Haible
openat: Fix warning.
* lib/openat
On Linux/glibc systems, I get this warning:
test-pty.c:24: warning: initialization from incompatible pointer type
test-pty.c:26: warning: initialization from incompatible pointer type
It's because glibc defines both forkpty and openpty with 'const' in the
last two parameters.
And also:
test-get
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Bruno Haible on 12/24/2009 11:20 AM:
> Eric Blake wrote:
>> I'm pushing this.
>
> I get this gcc warning, on Linux/glibc:
>
> test-nanosleep.c:24: warning: 'struct timepec' declared inside parameter list
> test-nanosleep.c:24: warning: i
Jim Meyering writes:
> -# define SET_BINARY(fd) /* nothing */
> +# define SET_BINARY(fd) 0
For me, with -Wall, plain 0 as a statement provokes a warning
"statement with no effect", but writing (void) 0 suppresses that
warning.
--
Ben Pfaff
http://benpfaff.org
Eric Blake wrote:
> I'm pushing this.
I get this gcc warning, on Linux/glibc:
test-nanosleep.c:24: warning: 'struct timepec' declared inside parameter list
test-nanosleep.c:24: warning: its scope is only this definition or declaration,
which is probably not what you want
test-nanosleep.c:24: war
Hi Bruno,
Without the patch below, a use like this:
if (expr)
SET_BINARY (fd);
would evoke a warning from gcc on most systems.
Ok to commit?
>From 9d0a1847b9288fb45b244cef1f1e1cceddaeebe5 Mon Sep 17 00:00:00 2001
From: Jim Meyering
Date: Thu, 24 Dec 2009 19:09:58 +0100
Subject: [PATCH]
Eric Blake wrote:
> --- a/lib/unistd.in.h
> +++ b/lib/unistd.in.h
> @@ -33,10 +33,17 @@
> #include
>
> /* mingw doesn't define the SEEK_* or *_FILENO macros in . */
> -#if !(defined SEEK_CUR && defined SEEK_END && defined SEEK_SET)
> +/* Cygwin 1.7.1 declares symlinkat in , not in . */
> +#if
Eric Blake wrote:
> These were the only includes that I found that were conditional on a
> particular module being in use.
Tricky stuff. There was a typo. Correcting it:
2009-12-24 Bruno Haible
* lib/stdlib.in.h (includes): Fix typo in today's commit.
--- lib/stdlib.in.h.orig
Eric Blake wrote:
> According to Eric Blake on 12/23/2009 2:45 PM:
>>> Possibility 1: Put the SIGNATURE_CHECK into a file tests/signature.h, and
>>> put
>>> ASSERT, SIZEOF and a few others into tests/macros.h. A test looks like this:
>>>
>>
>> It's less work for me if we go with possibility 2, but
Eric Blake wrote:
> I've noticed that several modules use link-warning, but only conditionally
> build the replacement header. The simplest example of this is isblank and
> ctype.in.h - since the only thing in that file that used GL_LINK_WARNING
> was isblank, but glibc includes isblank, the repla
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
I've noticed that several modules use link-warning, but only conditionally
build the replacement header. The simplest example of this is isblank and
ctype.in.h - since the only thing in that file that used GL_LINK_WARNING
was isblank, but glibc includ
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Eric Blake on 12/23/2009 2:45 PM:
>> Possibility 1: Put the SIGNATURE_CHECK into a file tests/signature.h, and put
>> ASSERT, SIZEOF and a few others into tests/macros.h. A test looks like this:
>>
>
> It's less work for me if we go with
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Eric Blake on 12/23/2009 9:17 PM:
> Hmm, looking at it further, it looks like we need to make all of our
> conditional #includes in the *.in.h files also be activated if
> GNULIB_POSIXCHECK is defined. I'll work on a followup patch for th
As a preparation for the ASSERT and SIGNATURE_CHECK changes, I'm applying the
common idiom to more test files: include the specification header first,
right after . This verifies that the header is standalone.
2009-12-24 Bruno Haible
* tests/test-argv-iter.c: Include header file being
25 matches
Mail list logo
