Re: error, verror: Don't call va_end twice.

Hi Bruno, Bruno Haible writes: > AFAICS, glibc has this bug already since 2001. Can you please file a glibc > bug for it? Sure. Reported here [1]. And patch here [2]. Collin [1] https://sourceware.org/bugzilla/show_bug.cgi?id=32233 [2] https://sourceware.org/pipermail/libc-alpha/2024-October/

Re: error, verror: Don't call va_end twice.

Collin Funk wrote: > I noticed that verror/error call va_end twice. Once in error_tail and > then again in the calling function. Thanks! Yes, it is wrong to invoke va_end (args); inside error_tail, because ISO C 23 § 7.16.1 says (emphasis is mine): "Each invocation of the va_start and va_cop

error, verror: Don't call va_end twice.

Hi Paul, I noticed that verror/error call va_end twice. Once in error_tail and then again in the calling function. This seems to have occurred when privatizing functions in glibc a while ago. I've pushed this patch to Gnulib. Only a one line change, but I don't have copyright papers for glibc, s

[PATCH 1/2] dirent: define DT_* macros on all platforms

* lib/dirent.in.h (DT_UNKNOWN, DT_FIFO, DT_CHR, DT_DIR, DT_BLK) (DT_REG, DT_LNK, DT_SOCK, DT_WHT): Define these on all platforms, if the system does not already define them. Check that they have distinct values. (_GL_DIRENT_S_ISWHT, _GL_DIRENT_S_IFWHT) [!(IFTODT && DTTOIF)]: New macros. (IFTODT, D

[PATCH 2/2] file-has-acl: no need for struct stat

Change the API of the new file_has_aclinfo function so that it no longer needs a struct stat *. In some cases this can help GNU ls avoid an unnecessary ‘stat’ call for each file it lists, which can be a significant win. * lib/acl.h (ACL_SYMLINK_NOFOLLOW): Change from 1 to UCHAR_MAX+1 so that it ca

gettext 0.22.5a Windows binaries released

I've just published the gettext 0.22.5a (and iconv 1.17) executables for Windows [1]. @Bruno: thank you for your valuable support! (And thanks to SignPath most of the binaries are now signed :P ) -- Michele [1] https://mlocati.github.io/articles/gettext-iconv-windows.html

Re: Building from the git repository fails with "alias_table defined but not used"

> On 02/10/2024 01:54, Yuri Kanivetsky wrote: > > Hi, > > > > https://gist.github.com/x-yuri/883efd92dff6a1ee77b14cff41c4fb12 This has: CC lib/libcoreutils_a-localcharset.o lib/localcharset.c:93:33: error: 'alias_table' defined but not used [-Werror=unused-const-variable=] 93 | stati

Re: Building from the git repository fails with "alias_table defined but not used"

On 02/10/2024 01:54, Yuri Kanivetsky wrote: Hi, https://gist.github.com/x-yuri/883efd92dff6a1ee77b14cff41c4fb12 I understand now that to make it build I need to configure with --enable-gcc-warnings=no. But is this expected or am I doing something wrong? I mean, shouldn't it build with --enable-

Re: Adapting changes for MSYS2?

Following up to > I wrote: > > MSYS, last released in 2016 [0], was a project with big hacks. For example, > > in the 'exec' system call, it replaces an argv[i] = "/dev/null" with "nul" - > > MSYS2 does similar things, albeit i

Re: [PATCH] file-has-acl: new function file_has_aclinfo

Paul Eggert wrote: > > ../../gllib/file-has-acl.c:427:21: error: use of undeclared identifier > > 'acl_extended_file' > >? acl_extended_file > > Thanks, fixed by installing the attached. Thanks. The Gnulib CI build is now green again. Bruno