pselect: Disable signature check on AIX.

--extract-recursive-dependencies pselect | grep extension extensions extensions-aix Collin >From 8ec056e11a5e0f4bb1aaffb623ecb5b7aebac757 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Fri, 25 Apr 2025 18:02:51 -0700 Subject: [PATCH] pselect: Disable signature check on AIX. * tests/test

Re: vasnprintf: Consider the grouping rule

Bruno Haible writes: > Thanks a lot for this report! Without you, I would not have noticed, > possibly not even before the next libunistring release. Glad to help. Just curious, do you do a libunistring release roughly around each new Unicode standard release? >> And copy nl_langinfo (THOUSEP)

Re: MinGW-w64 fenv changes.

Bruno Haible writes: > When working on the floating-point code, one needs to have the hardware > specification handy. In this case, it is the > Intel® 64 and IA-32 Architectures Software Developer’s Manual > Combined Volumes: 1, 2A, 2B, 2C, 2D, 3A, 3B, 3C, 3D and 4 > (4700 pages) > > Then,

MinGW-w64 fenv changes.

Hi Bruno, I was experimenting with Windows recently after the recent gperf thread. In a "MSYS2 MINGW64" window I found that a gnulib testdir of all modules fails to build due to the following assertion in fenv-env.c: verify (sizeof (fenv_t) >= sizeof (x86_387_fenv_t)); Upon investigating, I

Re: fbufmode: Don't require peeking into the FILE structure on recent Haiku

Bruno Haible via Gnulib discussion list writes: > With this patch and the recent fixes on the Haiku side, gnulib should no > longer > need to peek into the FILE structure on recent Haiku versions. Thanks! I was going to check if something similar worked when I had a chance. Glad it does.

tsearch tests: Disable tfind signature check on Cygwin.

ll cause annoying -Wdiscarded-qualifiers warnings in applications that expect the POSIX prototype [1]. Pushed the attached patch to disable the signature check and document the differing prototype. Collin [1] https://cygwin.com/pipermail/cygwin/2025-April/258032.html >From 30c887298af4a59915c98

random_r tests: Disable signature check on Haiku.

t random_data *)); | ^~~ This different prototype was documented in 2017. I've pushed the attach patch disabling the signature check and updating the documentation to acknowledge current Haiku has this prototype. Collin >From bdb1601d379c56a915c439fa66adbb64f0c

strsignal tests: Disable signature check on Haiku.

missing const is worth a workaround, so I have disabled the signature check and documented the bug. Reported to Haiku in case they want to comply with POSIX [1]. Collin [1] https://dev.haiku-os.org/ticket/19537 >From b05d9ebbc823972f568107c4ed9b79c6dfae496c Mon Sep 17 00:00:00 2001 From:

Re: Android API 8 compilation errors

ve pushed the attached patch to fix it. Collin >From d3d50d47e15b597f8a9a804bf81877bf842c60af Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Mon, 21 Apr 2025 20:08:30 -0700 Subject: [PATCH] getloadavg: Fix typo in previous commit. * lib/getloadavg.c (getloadavg): Remove '__' prefix from 'defined'.

Re: vasnprintf: Consider the grouping rule

Hi Bruno, Bruno Haible via Gnulib discussion list writes: > This patch fixes it, by considering the 'grouping' sequence of numbers. On FreeBSD 15.0 (cfarm427) I see the following warnings: In file included from unistdio/u16-u16-vasnprintf.c:57: ./vasnprintf.c:5224:77: warning: incompat

Re: C++ std::

Hi Bruno, Bruno Haible writes: > I try to avoid the C++ Standard Library, because Thanks for the detailed response. I certainly agree regarding the unreadability of the C++ standard library. Collin

Re: Update users.txt

Bruno Haible via Gnulib discussion list writes: > GNU gperf now uses Gnulib: A 30% speedup of gperf is realized through the > gnulib module 'hash-map'. Nice improvement and congrats on the new release. Out of curiosity though, why not use std::unordered_map [1]? Is C++ standard library not port

Re: [PATCH 2/5] Pacify GCC 15’s new -Wtrailing-whitespace option

ut now test -n "$iconv_arg1" will always be true. Bruno, I have attached a patch to adjust libiconv to this change. Also, another patch since ACLOCAL was updated in Makefile.devel to aclocal-1.17. But libcharset/Makefile.devel had it set to aclocal-1.16. I assume this was unintentional, fee

Re: vasnprintf tests: Add a test case that showcases a Solaris bug

Bruno Haible via Gnulib discussion list writes: > In printf implementations, it is easy to miss the fact that for %.50g > the implementation needs to allocate room for the thousands-separators. > I checked various systems, and Solaris printf() was found to crash in > such circumstances. Good cat

Re: immutable: Avoid redefining macros on AIX.

warning. > How about this? > > /* AIX defines a macro ALIGNMENT in . Undefine it. */ That description is more accurate. Applied the attached patches with your suggestions. Thanks! Collin >From 3827476ac89e400a2383107a2c1919e36970f486 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sa

stdlib: Avoid syntax errors in libstdc++ header files on AIX.

l, strtoul, etc. This is the same as Bruno found on Solaris 11 OmniOS previously [1]. The same fix works in this case. Pushed the attached patch to fix it. [1] https://lists.gnu.org/archive/html/bug-gnulib/2024-07/msg00129.html >From 79fa09434c01cae7ee72bc1534257296818e0143 Mon Sep 17 00:00

immutable: Avoid redefining macros on AIX.

s by default. I have pushed the attach two patches fixing this, and a similar one in the aligned-malloc tests. Collin [1] https://www.open-std.org/jtc1/sc22/wg14/www/docs/n3220.pdf >From 83a07409b06600750ad194aed8989ae1df3f6140 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sat, 12 Apr 2025

Re: stddef-h: Make a configure test work with upcoming GCC 15

0 2001 From: Collin Funk Date: Tue, 8 Apr 2025 21:20:39 -0700 Subject: [PATCH] stddef-h: Silence autoconf warnings introduced in previous commit. * m4/stddef_h.m4 (gl_STDDEF_H): Use AC_LANG_SOURCE inside call to AC_COMPILE_IFELSE. --- ChangeLog | 6 ++ m4/stddef_h.m4 | 7 --- 2

Re: Haiku fixes for for fseeko, fbufmode, fflush, and fseterr.

Hi Bruno, Bruno Haible writes: > Thanks a lot for this. Let me update some comments accordingly. Thanks! On the Haiku bug report, your input might be appreciated [1]. They want to avoid issues if they break internal ABI. I suggested adding __fbufmode, even though fbufmode is a Gnulib function

Haiku fixes for for fseeko, fbufmode, fflush, and fseterr.

ollin [1] https://lists.gnu.org/archive/html/bug-gnulib/2025-03/msg00051.html [2] https://dev.haiku-os.org/ticket/19479 >From 58195177985347cff52d402622d264daa2f8e179 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sun, 6 Apr 2025 19:19:35 -0700 Subject: [PATCH 1/4] fseeko: Port to recent Haiku

Re: [PATCH] gnulib-tool.py: Fix missing space.

Hi Eric, Eric Blake writes: > Noticed this while working on m4. Pushing as obvious, since it is a > difference between the .sh and .py versions. I'm not sure if it is > worth trying to enhance the gnulib-tool testsuite to add coverage for > error messages like this. Thanks! I don't think a t

mountlist tests: Fix link error. (regr. 2025-03-18).

iled with exit code 1 (use -v to see invocation) And a similar error will probably happen on non-glibc systems. Collin >From f1daedcb4c6c653dfc2936847dfe55f5a076d94b Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sun, 23 Mar 2025 20:32:01 -0700 Subject: [PATCH] mountlist tests: Fix lin

doc: Document NetBSD acl functions.

#x27;. So I think it should be fine. Collin >From 01c06a1aadcb23dee09189504a02825dd1092c42 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Fri, 28 Mar 2025 14:58:31 -0700 Subject: [PATCH] doc: Document NetBSD acl functions. * doc/acl-resources.txt: Document that NetBSD supports all AC

Re: Silence some -Wunterminated-string-initialization warnings

Hi Bruno, Patch looks good, thanks! Bruno Haible via Gnulib discussion list writes: > +static const char b64c[64] _GL_ATTRIBUTE_NONSTRING = >"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; I didn't know about __attribute__ ((__nonstring__)) until this patch. That looks

Re: crypto/gc: Silence some -Wcalloc-transposed-args warnings

Bruno Haible via Gnulib discussion list writes: > Since the first argument of calloc() is meant to be the number of elements > to allocate and the second argument is meant to be the size of each element, > and since here each element is a _gc_cipher_ctx or _gc_hash_ctx, the > warning makes sense.

Re: doc: Document NetBSD acl functions.

Hi Bruno, Bruno Haible writes: > Yup. But the comments can be updated. Done as follows: Thanks! Looks like it was just the comments that needed updating. Collin

newlocale: Fix implicit function declarations.

ed. I have pushed the attached patch to fix them. Collin >From 7d0d2480143d46dc9ee8dfd17d4149b84d5eab2e Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sun, 23 Mar 2025 14:33:49 -0700 Subject: [PATCH] newlocale: Fix implicit function declarations. * lib/newlocale.c [HAVE_NEWLOCALE]: Include s

Re: [PATCH] maintainer-makefile: Check spelling using 'codespell'.

Hi Simon, Simon Josefsson via Gnulib discussion list writes: > I've been using 'codespell' to find typos in several projects for a long > time, via ad-hoc syntax-check rules in cfg.mk. I tried to clean up the > rule and make it generic enough to be useful everywhere. I've pushed > the attached

Re: [PATCH] maintainer-makefile: Check spelling using 'codespell'.

Simon Josefsson writes: > I think all of that could be converted into cfg.mk variables and use my > 'make syntax-check' codespell rule. But it seems Emacs isn't using > 'make syntax-check' from gnulib. Sorry for not being clear. Emacs doesn't use maintainer-makefile and such. I was just mention

Re: new module 'gocomp-script'

Hi Bruno, Bruno Haible via Gnulib discussion list writes: > +# Avoid error "CC environment variable is relative; must be absolute path" > +# on native Windows. Cf. > +# > > +unset CC > +unset CXX The CC and CXX vari

utimens: Fix missing declarations on NetBSD (regr. 2024-09-16).

s://github.com/NetBSD/src/commit/ff84366ca60c698af8d1ea72aa3871bee0ba8463#diff-aa03d0529d55ab3fffd58831b20dc2c1742770e2d82efcf5c3e9843bde19adb4 >From 39db933c36305612b79300aa34f4d0d926816cd3 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sun, 23 Mar 2025 16:36:26 -0700 Subject: [PATCH] utimens: Fix

Re: mountlist for native Windows

Bruno Haible writes: > And with this patch, mount points other than the 26 drives are listed. > For example, I created a separate disk, formatted it as NTFS, mounted > it at S:\, added some files, unmounted it from S:\, and finally > mounted it at C:\Users\bruno\mt_scratch. > > The test program n

futimens: Work around a GNU/Hurd bug.

how_bug.cgi?id=32803 [3] https://lists.gnu.org/archive/html/bug-gnulib/2024-09/msg00120.html >From 23a3940f61f46e84067c4b472920e1c7d34459c2 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Tue, 18 Mar 2025 21:50:00 -0700 Subject: [PATCH] futimens: Work around a GNU/Hurd bug. * m4/futimens.m4 (gl_

Re: mountlist for native Windows

mation. But I made it fail if no mount points are returned (NULL). Hopefully that will catch any systems that haven't been covered yet. I pushed both attached patches. Collin >From 5f693944c4a5d9ad8334c27752a0a286d3fcb273 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Tue, 18 Mar 20

Prefer the __gnu_hurd__ macro over __GNU__.

ing some manual adjustments to lib/getloadavg.c. Collin >From 43fccde48cbd878ea7b155fc764c1e99a23458d6 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Tue, 18 Mar 2025 18:37:48 -0700 Subject: [PATCH] Prefer the __gnu_hurd__ macro over __GNU__. * lib/get_ppid_of.c: Replace use of __GNU__ with

utimensat: Make sure exit status in configure check doesn't exceed 127.

8066f870c7e7fac65a4229db5f21 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Tue, 18 Mar 2025 18:12:39 -0700 Subject: [PATCH] utimensat: Make sure exit status in configure check doesn't exceed 127. Reported by Bruno Haible in <https://lists.gnu.org/archive/html/bug-gnulib/2025-03/msg00061

Re: utimensat: Work around a GNU/Hurd bug.

Forgot that m4 serial numbers exist. Incrementing it just in case it matters to anyone. >From 4f98832331353fc1a599d1b7fdc89a817da23760 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Mon, 17 Mar 2025 22:13:04 -0700 Subject: [PATCH] utimensat: Increment serial number for previous commit. *

utimensat: Work around a GNU/Hurd bug.

urd component of glibc [1]. And pushed the attached patch to fix it in Gnulib. Collin [1] https://sourceware.org/bugzilla/show_bug.cgi?id=32802 >From 381815da7f8d319e7dd9a59509f901fc9948918c Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Mon, 17 Mar 2025 21:16:22 -0700 Subject: [PATCH] utime

Re: m4/mountlist.m4: remove AC_MSG_ERROR from gl_MOUNTLIST?

quot; exist, but it reads to me like they are just aliases for mounts with drive prefixes [1]. Collin [1] https://learn.microsoft.com/en-us/windows/win32/fileio/volume-mount-points >From 95bc393f6dc6586906b094ecb267b75b106c2925 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Mon, 17

Re: m4/mountlist.m4: remove AC_MSG_ERROR from gl_MOUNTLIST?

Hi Bruno, Bruno Haible via Gnulib discussion list writes: > But it is equally unusual for a Gnulib module to provide just a dummy stub > on some platforms. > > Therefore, ideally, someone would port this module to native Windows > (considering drive letters, JOINs, SUBSTs, and \\...\ shares). Th

Recent Haiku update breaks stdio_ext.h functions.

A recent Haiku commit breaks the stdio_ext.h functions [1]. I discovered this in a testdir of all modules where I see this: gcc -DHAVE_CONFIG_H -DEXEEXT=\"\" -DEXEEXT=\"\" -DNO_XMALLOC -DEXEEXT=\"\" -I. -I.. -DGNULIB_STRICT_CHECKING=1 -fvisibility=hidden -g -O2 -MT fbufmode.o -MD -MP -MF $

vma-iter: Detect executable memory segments on Haiku (regr. 2011-01-25).

s://github.com/haiku/haiku/blame/b989960b81e5826d54acad82a2c4a3e685984f8a/src/system/libroot/posix/sys/mman.cpp#L132 >From 5077f67040ab62a3ea1335656ac0b6f76bfbea66 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Thu, 13 Mar 2025 19:11:08 -0700 Subject: [PATCH] vma-iter: Detect executable memory

dup3: Fix behavior for equal file descriptors on Haiku.

rom ed91d9b1c2f697a39116da46b553dfc89474f490 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Wed, 12 Mar 2025 18:55:26 -0700 Subject: [PATCH] dup3: Fix behavior for equal file descriptors on Haiku. * lib/dup3.c (dup3) [__HAIKU__]: Set errno to EINVAL and return -1 if both file descriptors are equal. * doc/posix-functions/dup3.t

Re: unistd-h: Make sure O_CLOEXEC is defined.

Hi Bruno, Bruno Haible writes: > The cause is the recursive include -> -> , > that was introduced on 2025-02-16 and that has the effect that fcntl.h > examines the value of O_NONBLOCKING before this macro has gotten its > final value. Thanks for the fix. I rarely test Haiku since, in my expe

Re: Haiku

Bruno Haible writes: > Find attached the newest set of test failures, in case you have time > to report issues to their bug tracker. Thanks! I'll have a look at them and see if I can bring the number of failures down a bit. Collin

Re: getaddrinfo: Fix AI_NUMERICSERV configure check on GCC 14 (regr. 2025-02-12).

Collin Funk writes: > If I remember correctly, this code would warn on previous versions of > GCC. But on GCC 14 it is an error by default. Forgot to attach the patch. Collin >From 7029d554682583586d56deefb60af7a1aae233ef Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Tue, 4 Mar 2

getaddrinfo: Fix AI_NUMERICSERV configure check on GCC 14 (regr. 2025-02-12).

On GNU/Linux with GCC 14.2, I noticed the following: $ gnulib-tool --create-testdir --dir testdir1 getaddrinfo $ cd testdir1 $ ./configure [...] checking for getaddrinfo... yes checking whether getaddrinfo supports AI_NUMERICSERV... no [...] In config.log there is the

Re: new module 'vc-mtime'

Bruno Haible via Gnulib discussion list writes: > unsigned long git_log_time; > if (xstrtoul (line, &endptr, 10, &git_log_time, NULL) == LONGINT_OK > && endptr == line + strlen (line)) > { > mtime->tv_sec = git_log_time; >

Re: unistd-h: Make sure O_CLOEXEC is defined.

changes. Collin >From 5662b5eef083b688626d8b502290b886413d635d Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sun, 16 Feb 2025 13:58:35 -0800 Subject: [PATCH] doc: Adjust platforms that define O_CLOEXEC in unistd.h. Suggested by Bruno Haible. * posix-headers/unistd.texi: Mention that macOS 12 and AIX 7.3 do not defi

unistd-h: Make sure O_CLOEXEC is defined.

https://austingroupbugs.net/view.php?id=1906 [3] https://sourceware.org/bugzilla/show_bug.cgi?id=32706 >From 10331a8ba5e9548671907a09eb6d5d7260543c39 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sun, 16 Feb 2025 12:23:57 -0800 Subject: [PATCH 1/2] unistd-h: Make sure O_CLOEXEC is defined. * modules/unistd-h

acl-permissions, file-has-acl: Depend on stdint-h for SIZE_MAX.

le-has-acl Collin >From c1a7af9bc8ebc9bbe9fb8d62664103810a76ba10 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Tue, 11 Feb 2025 18:44:33 -0800 Subject: [PATCH] acl-permissions, file-has-acl: Depend on stdint-h for SIZE_MAX. * lib/acl-internal.h: Include stdint.h. (SIZE_MAX): Remove defi

Re: [PATCH] memalignment: document

deda30f77a1779e4d0da3c2b8f688b Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sat, 8 Feb 2025 17:40:48 -0800 Subject: [PATCH] doc: Document free_aligned_sized and free_sized added in C23. * doc/posix-functions/free_aligned_sized.texi: New file. * doc/posix-functions/free_sized.texi: New file. *

Re: announce-gen project link and mirrors

Simon Josefsson via Gnulib discussion list writes: >> Here are the compressed sources: >> https://ftp.gnu.org/gnu/diffutils/diffutils-3.11.tar.gz (3.3MB) >> https://ftp.gnu.org/gnu/diffutils/diffutils-3.11.tar.xz (1.9MB) >> >> Here are the GPG detached signatures: >> https://ftp.gnu.org

Re: encoding of "C" locale

Bruno Haible writes: >> I was thinking maybe adding a way to force ASCII on Haiku, but that >> seems to hacky. > > How would you want to do that? If the system has no locale with ASCII > or ISO-8859-1 encoding (such as on macOS, Haiku, and Android), you can't > force it. You're right. That was a

Re: test-xfail: Define a condition for Haiku

Hi Bruno, Bruno Haible via Gnulib discussion list writes: > Some portability problems are specific to Haiku. Therefore it makes sense > to mark some test failures as "expected failure on Haiku only". Makes sense, thanks. Somewhat related, I recently tried Haiku for the first time and noticed t

Re: gnulib-tool: Allow compiler warnings in Gnulib code

Hi Bruno, Bruno Haible via Gnulib discussion list writes: > So, these compilation units in coreutils MUST be compiled with -Wno-error. > > This patch achieves that. > > Previously we had added -Wno-error only on the gnulib tests. This patch > adds it also for the source files in lib/. Thanks, t

doc: Document version-etc, version-etc, and argp-version-etc.

you understand how they work. Collin >From 64b120efb9cad480983cb779e4fc552e5fd79230 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Tue, 28 Jan 2025 20:52:17 -0800 Subject: [PATCH] doc: Document version-etc, version-etc, and argp-version-etc. * doc/version-etc.texi: New file. * doc/gnulib.texi (

Re: OmniOS strerrorname_np

Collin Funk writes: > I was going to check OmniOS again anyways since on my virtual machine > test-ptsname_r failed (it is skipped on CI). > > It is because OmniOS will fail and put a NUL byte at the first character > in the buffer. So hopefully that should be an easy fix as well.

strerrorname_np: Work around a bug on Solaris 11 OmniOS.

Collin Funk writes: >> ../../build-aux/test-driver: line 121: 54: Memory fault(coredump) >> FAIL test-strerrorname_np (exit status: 267) >> >> Do you have time to investigate this? > > Sure, I will take a look at it later today. As promised, I have pushed the atta

Re: OmniOS strerrorname_np

Hi Bruno, Bruno Haible writes: > FAIL: test-strerrorname_np > == > > ../../build-aux/test-driver: line 121: 54: Memory fault(coredump) > FAIL test-strerrorname_np (exit status: 267) > > Do you have time to investigate this? Sure, I will take a look at it later today. I

Re: tests: Fix a few compilation errors on Solaris 11 OmniOS with gcc 14

Hi Bruno, Bruno Haible via Gnulib discussion list writes: > The CI, now with gcc 14, reports a compilation error on Solaris 11 OmniOS: I saw this yesterday as well but didn't have the time to complete a fix. If you look in you will see something like the following: #if defined _XPG4_2

Re: char-h-c23: Fix compilation error on OmniOS.

Collin Funk writes: > $ gcc --version > gcc (OmniOS 151052/14.2.0-il-1) 14.2.0 I was wondering why this failure didn't occur in the GitHub CI. It is because the CI uses GCC 13 where it is a warning. In GCC 14 they made it an error by default (-Werror=incompatible-pointe

char-h-c23: Fix compilation error on OmniOS.

t is of type 'char **' 47 | extern size_t iconv(iconv_t, const char **_RESTRICT_KYWD, *** Error code 1 make: Fatal error: Command failed for target `lc-charset-unicode.o' I've pushed the attached patch to cast using the ICONV_CONST macro, which allows it to

Re: renameatu: Work around a GNU/Hurd bug.

Hi Bruno, Bruno Haible writes: > Thanks. Let me tweak the configure test: > - It is confusing that the test has an open file descriptor to the file > being renamed. > - It is useful for the test to return a bit mask, instead of having > two 'return 1;' cases. Namely, if the test fail

renameatu: Work around a GNU/Hurd bug.

g.cgi?id=32570 >From 1e3de03067bf1ca2cec2dc2ce64f682bcf66b5d9 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sat, 18 Jan 2025 21:06:49 -0800 Subject: [PATCH] renameatu: Work around a GNU/Hurd bug. * m4/renameat.m4 (gl_FUNC_RENAMEAT): Check if renameat2 handles trailing slashes. * doc/glibc-f

Re: linkat tests: Avoid failure on GNU/Hurd.

et fixed [1]. In the meantime, I have pushed the attached patch restoring the test to its original state. And documented the bug so one, after manually testing, can see the failure is expected. Collin [1] https://sourceware.org/bugzilla/show_bug.cgi?id=32569 >From 82c416d7aec0a8bd52fa8c3

linkat tests: Avoid failure on GNU/Hurd.

link to a directory. Once we allow errno == EINVAL for this case the tests pass. I've just gone ahead and pushed the attached since I don't think the errno is worth correcting in this case. Collin >From 9f065777c811130d9b6b539dbbedc9924d6f7916 Mon Sep 17 00:00:00 2001 From: Collin Funk

libgmp-mpz: Respect Automake's silent-rules.

Collin Funk writes: > In Coreutils, which uses --enable-silent-rules by default, I see: > > $ make > GEN .version > GEN lib/alloca.h > GEN lib/arpa/inet.h > GEN lib/configmake.h > if test -n 'gcc'; then \ >

Re: gnulib makefile snippets

Bruno Haible via Gnulib discussion list writes: >> Maybe simple to add a 'Files: etc/make/git-version-gen.mk' to >> modules/git-version-gen? >> >> Although I suspect gnulib-tool will have to be teached about a >> --etc-base then. > > Not 'etc/' please. One should think three times before naming

Re: strerrorname_np on GNU/Hurd

Hi Bruno, Bruno Haible writes: >> How about I #if 0 these checks out in the tests and Gnulib replacement >> function for now so the behaviors are the same? > > I would only do so if there was no hope that the glibc bug ever gets fixed. > But since > - it was fixed for other glibc ports, > -

strerrorname_np on GNU/Hurd

Hi Bruno, I noticed that test-strerrorname_np.c fails on the following line on GNU/Hurd: /* GNU/Hurd */ #if defined ED_ALREADY_OPEN ASSERT (strcmp (strerrorname_np (ED_ALREADY_OPEN), "ED_ALREADY_OPEN") == 0); #endif This is because glibc doesn't have GNU/Hurd errors in sysdeps/gn

Re: stdlib-h: Define WCOREDUMP, as required by POSIX:2024

Bruno Haible writes: > The CI reports a compilation error on all glibc systems. Thanks for the fix and glibc bug report. I forgot that stdlib.h is supposed to define these macros as well. Collin

crc: Respect Automake's silent-rules.

rpa/inet.h GEN lib/configmake.h GEN lib/crc-sliceby8.h GEN lib/ctype.h Collin >From 539fa8c1d705c63d4e034def4719b0977d8a72a2 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sun, 12 Jan 2025 22:28:44 -0800 Subject: [PATCH] crc: Respect Automake's silent-r

Re: sys_wait-h: Update comment.

Bruno Haible writes: > Thanks. Let me update the documentation accordingly: Thanks! I also see you marked it "DONE" in the maint-tools repository. I was debating doing it but was unsure since wait() and waitid() are not implemented. But I guess it makes sense because those are existing interfac

sys_wait-h: Update comment.

anged, 7 insertions(+), 1 deletion(-) diff --git a/ChangeLog b/ChangeLog index d33a584ad8..99f913f1e5 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,3 +1,9 @@ +2025-01-10 Collin Funk + + sys_wait-h: Update comment. + * lib/sys_wait.in.h (WCOREDUMP): Remove comment about this macro being + non-stan

Re: servent: Add tests.

Oops, that is embarrassing... Fixed with your recommendation. Collin >From dd8057e53a8b4e710528d35948ca07220292b931 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Mon, 6 Jan 2025 18:28:35 -0800 Subject: [PATCH] servent tests: Fix mistake in previous commit. Reported by Bruno Haible. * te

Re: servent: Add tests.

Hi Bruno, Bruno Haible writes: > But should this condition > > #if defined _WIN64 && !defined __CYGWIN__ > > not better be > > #if !(defined _WIN32 && !defined _WIN64 && !defined __CYGWIN__) > > ? Good question. I wasn't exactly sure how _WIN32 and _WIN64 were defined and copied the conditi

Re: servent: Add tests.

x27;t see it used as a function pointer and 32-bit Windows doesn't seem worth the effort. Pushed the attached patch to disable the tests there and document the declaration. Collin >From 71fbdb50ab5b0da7053a93975612b42f0d2e591b Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Mon, 6 Jan 2025

Re: new modules mbs_startswith, mbs_endswith

Hi Bruno, Bruno Haible via Gnulib discussion list writes: > I'm not opposed to using 'bool'. But when I saw that no function in > gnulib's , , or , so far uses 'bool', > it made me hesitate. > > Should we break the "tradition" here to use 'int' for a Boolean value, > and actually use 'bool' for

Re: servent: Add tests.

c we just disable the signature test. What is the correct option in this case? Thanks, Collin >From 020ebd643d58153f1fbab8faa90fb09899950bc4 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Tue, 31 Dec 2024 19:43:28 -0800 Subject: [PATCH 1/2] servent: Make sure netdb.h is generated. * modules

Re: servent: Add tests.

Forgot to add attachment... >From e9c1d94f58eaacee919bb2015da490b980a5eedf Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Mon, 30 Dec 2024 19:00:01 -0800 Subject: [PATCH] servent: Add tests. * tests/test-servent.c: New file. * modules/servent-tests: New file. --- ChangeLog |

servent: Add tests.

I added a simple test for the servent module. All it does is check that the function prototypes of getservbyname and getservbyport are correct and that the functions can be called. Therefore it will fail if $(SERVENT_LIB) is set incorrectly. POSIX says [1]: This information [serivces] is cons

signal-h tests: Check that SIG2STR_MAX is properly defined.

Since Gnulib's signal.h defines SIG2STR_MAX like POSIX requires, the test should check for it. Done with the attached patch. Collin >From aaf13948de429369aa71e7bf9d6d05196cba4ebf Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sun, 29 Dec 2024 11:56:28 -0800 Subject: [PATCH] signal

sig2str tests: Add signature check.

Now that sig2str and str2sig are a part of POSIX we should add signature checks so we can find out if they are incorrect on any platforms. Done through the attached patch. Collin >From bf6f7d8a81f231b14ab50194a46bbcbca21835b1 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sun, 29 Dec 2024

Re: cgit .tar.gz code snapshots

Hi, fosslinux writes: > Without going into our technical structure - yes and no. However, we > are actively moving away from dynamically generated tarballs (to > making them ourselves from Git clones), not just for gnulib, but for > everything where we consume a Git snapshot. We can get past thi

Re: doc: Mention the servent module.

at. :) Collin >From 1d1888d09f0f67517cdc4ce87e2136e0c0ac8941 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sat, 28 Dec 2024 18:11:03 -0800 Subject: [PATCH] doc: Add missing module index. Suggested by Bruno Haible in: <https://lists.gnu.org/archive/html/bug-gnulib/2024-12/msg00240.html>. * doc/posi

Re: doc: Mention the servent module.

3f2b425e76ea826228d1a0102fc965e3 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sat, 28 Dec 2024 17:26:28 -0800 Subject: [PATCH] doc: Fix the previous commit. Suggested by Bruno Haible in: <https://lists.gnu.org/archive/html/bug-gnulib/2024-12/msg00236.html>. * doc/posix-functions

doc: Mention the servent module.

ck2-getservbyname [2] https://learn.microsoft.com/en-us/windows/win32/api/winsock2/nf-winsock2-getservbyport >From 2d1c0b2f38c81d1c8eca2f06b39264f5029b97fe Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sat, 28 Dec 2024 16:19:01 -0800 Subject: [PATCH] doc: Mention the servent module.

Re: iconv_open: doesn't handle NetBSD bmake

Bruno Haible via Gnulib discussion list writes: > My conclusion is: The 'bmake' program that you tried is inferior > to what NetBSD is using, and since on GNU systems we have GNU make, > there is no need to support or even try another 'make' program. I agree. Also, Simon I remember Debian's bma

Re: crc-x86_64: Fix compilation error with clang

Hi Bruno, Bruno Haible via Gnulib discussion list writes: > I picked approach c), and committed the two attached patches. Tested both > without libtool (in a testdir) and with libtool (in some package that uses > gnulib for some shared library). Would __attribute__ ((target ("..."))) not work h

Re: base64: gcc-15 compile issue

Bruno Haible via Gnulib discussion list writes: > In this case, the code is perfectly fine: it initializes an array of > size 64 with 64 characters. There is no better way to write such an > initialization. If I interpret the warning correctly, I don't think it would trigger with something like

Re: Reducing duplication of bit operation code.

earn.microsoft.com/en-us/cpp/overview/compiler-versions?view=msvc-170#version-macros >From 1a9d52aadabd3862168da43926ab4518bd7ae02f Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Thu, 19 Dec 2024 17:53:48 -0800 Subject: [PATCH 1/2] stdbit-h: Detect MSVC __typeof__ support. * lib/stdbit.in.h

Re: Reducing duplication of bit operation code.

Hi Paul, Paul Eggert writes: > Why not instead encourage callers of these modules to use the > C23-style modules instead? For example, mark ffs as obsolescent and > suggest stdc_trailing_zeros as its replacement. This should improve > overall maintainability in the long run, while lessening oppo

Reducing duplication of bit operation code.

s and can do the rest if there isn't any disagreement. Collin >From abacaf822fe21d1acfc766b33fa1ba06867c39c6 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Tue, 17 Dec 2024 12:34:40 -0800 Subject: [PATCH] count-one-bits: Implement using stdbit.h. * lib/count-one-bits.h: Inc

Re: gnulib-tool.py: Optimize --extract-recursive-dependencies.

Hi Bruno, Bruno Haible writes: > Reviewing the new code now, I see a couple of details that can be improved: Thanks, your changes make things much cleaner. Collin

Re: gnulib-tool.py: Optimize --extract-recursive-dependencies.

Hi Bruno, Collin Funk writes: > Thanks, I reproduced your findings. I'll have a look at fixing it with > your suggestions. I've applied a patch following your suggestions. I believe the slowness in --extract-dependents caused by my previous patch was caused by the parsing

Re: gnulib-tool.py: Optimize --extract-recursive-dependencies.

anks, I reproduced your findings. I'll have a look at fixing it with your suggestions. Collin >From a67335ecf3ac57feeacb156d7871c88f8ea80817 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Wed, 4 Dec 2024 17:28:44 -0800 Subject: [PATCH] Fix a recent ChangeLog entry. --- ChangeLog | 2 +- 1 file

gnulib-tool.py: Optimize --extract-recursive-dependencies.

1ac78fd4b2a2d477 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Tue, 3 Dec 2024 19:17:52 -0800 Subject: [PATCH] gnulib-tool.py: Optimize --extract-recursive-dependencies. * pygnulib/GLModuleSystem.py (GLModuleSystem.list): Add optional argument to include test modules. (GLModule._getDepe

Re: announce-gen: Show an SHA256 sum that can be verified more easily

| 7 +++ build-aux/announce-gen | 45 +++--- 2 files changed, 45 insertions(+), 7 deletions(-) diff --git a/ChangeLog b/ChangeLog index 6d6fc99169..cc85aeb4b6 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,3 +1,10 @@ +2024-12-01 Collin Funk + +

Re: doc: Document that glob includes glob_pattern_p.

mary location). Oops, should have checked another file to see I was missing something. Thanks, I pushed the fix. Collin >From 5423426e996aa2b62b7a785384e7f125ee4ee7e6 Mon Sep 17 00:00:00 2001 From: Collin Funk Date: Sun, 24 Nov 2024 11:17:29 -0800 Subject: [PATCH] doc: Add missing module i

  1   2   3   4   5   6   7   8   >