> On May 16, 2017, at 8:12 AM, Eric Blake wrote:
>
>> On 05/16/2017 12:47 AM, Akash Rawal wrote:
>> Hello everyone
Hi,
>> I observed that build-aux/bootstrap uses $(command) for command
>> substitution which is known to be less portable than `command`.[1]
>
> It's less portable only if you st
The first fallout of the recent Coverity scan:
2017-05-16 Bruno Haible
javacomp: Fix handle leak.
Found by Coverity.
* lib/javacomp.c (get_classfile_version): Close fd before returning.
diff --git a/lib/javacomp.c b/lib/javacomp.c
index fe15b2c..3d7d457 100644
--- a/l
Although relocate() in practice does not produce memory leaks (in the
sense of _repeated_ loss of memory blocks), it is sufficiently often reported
by analysis tools (valgrind, coverity) that I'm now doing something against it.
2017-05-16 Bruno Haible
relocate: Make it easier to recla
On 05/16/2017 10:41 AM, Bruno Haible wrote:
Why not
# define FALLTHROUGH
? Would that generate warnings about empty statements?
Yes, that was the worry. I didn't actually observe the warnings, but I
assume some compilers will generate them.
Thanks for the sigpipe-tests fix.
Followup to this patch from 2014-12-09: A bit more consistent coding style:
- The argument 'pathname' must be non-NULL. No need to test it against NULL.
- Make use of ISSLASH macro consistently.
- For characters, I find a comparison with '\0' clearer than a boolean negation.
Limit boolean express
Hi Paul,
> +#ifndef FALLTHROUGH
> +# if __GNUC__ < 7
> +# define FALLTHROUGH ((void) 0)
Why not
# define FALLTHROUGH
? Would that generate warnings about empty statements?
> diff --git a/tests/test-sigpipe.c b/tests/test-sigpipe.c
> index 48b22f1..e75372d 100644
> --- a/tests/test-sigpipe.c
* build-aux/gcc-warning.spec:
* m4/manywarnings.m4 (gl_MANYWARN_ALL_GCC):
Add GCC 7 warnings, notably -Wimplicit-fallthrough=5, which
requires a non-comment fallthrough attribute. This is a bit
cleaner than the comment versions.
* lib/strftime.c, lib/dfa.c, lib/fnmatch.c, lib/mbrtowc.c:
* lib/vasn
On 05/16/2017 12:47 AM, Akash Rawal wrote:
> Hello everyone,
>
> I observed that build-aux/bootstrap uses $(command) for command
> substitution which is known to be less portable than `command`.[1]
It's less portable only if you still intend to develop on any system
with /bin/sh as a non-POSIX sh
Oh, just realized that ChangeLog has to be manually edited.
Here is the amended patch.
With Best Regards, Tim
On 05/16/2017 03:52 PM, Tim Rühsen wrote:
> Bruno ! You saved my day !
>
>
> I put your description with some small amendments into the README.
> Here's the patch.
>
>
> With Best
Hello everyone,
I observed that build-aux/bootstrap uses $(command) for command
substitution which is known to be less portable than `command`.[1]
Is it on purpose? Thanks in advance.
Regards,
Akash Rawal.
[1]
https://www.gnu.org/software/autoconf/manual/autoconf-2.63/html_node/Shell-Substi
Bruno ! You saved my day !
I put your description with some small amendments into the README.
Here's the patch.
With Best Regards, Tim
On 05/16/2017 03:17 PM, Bruno Haible wrote:
> Hi Tim,
>
>> Not sure how to build and run the test suite. The README just points me
>> to a .texi file which
Hi Tim,
> Not sure how to build and run the test suite. The README just points me
> to a .texi file which is hardly readable, at least I get a headache
> after ~10s.
I agree. The README should better say to
$ cd doc
$ make gnulib.html
$ xdg-open gnulib.html
or read the manual at
https://w
Hi,
I just git cloned gnulib to work on certain bugs.
Not sure how to build and run the test suite. The README just points me
to a .texi file which is hardly readable, at least I get a headache
after ~10s.
doc/README... something about web pages.
I tried 'make' - nothing happened.
'make che
13 matches
Mail list logo
