The new "std-gnu23" module has the stated goal of:
"...to update the c99 module to depend on std-gnu23
instead of on std-gnu11, and to make std-gnu11 obsolete."
in upstream commit 8990abb50 ("std-gnu23: new module").
However, for now, it's design is optional, so that
definitions of the latest standard module overrides the former.
At some point, upstream gnulib will replace the dependency
instead of add it alongside the older one.
Because all macros are copied to the aclocal directory,
for complex projects, not including the module
may cause the macros to apply only to some subdirectories
rather than all of them and top-level together.
For projects that import source from local gnulib,
always include the std-gnu23.m4 macros for consistency.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/19748
Signed-off-by: Robert Marko <robimarko@gmail.com>
After eliminating the possibility of automake having a bug
by testing a revert to the recent updates to automake,
the problems regarding autoreconf with some packages
was bisected to the gnulib update instead, through aclocal macros.
With the new module, std-gnu23, some packages are failing build
due to both the host compiler and cross compiler being tested for
availability of C23 standard features with the configure script.
The results of one is being cached and used for the other,
while the two compilers are different versions and may or may not
both support C23 options and would otherwise have conflicting results.
A similar patch may have to be done
for the next release of Autoconf
if upstream GNU does not accept this solution.
Reported-by: Georgi Valkov <gvalkov@gmail.com>
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/19627
Signed-off-by: Nick Hainke <vincent@systemli.org>
The latest versions of gettext rely on several changes to gnulib
including both changes to modules and new modules
and some previously gettext specific code being moved to gnulib.
Backport these changes in order to allow updating gettext
while using the local gnulib copy of sources.
Add patch:
- 640-mem-hash-map.patch
- 645-next-prime.patch
- 646-hashcode-string.patch
- 647-hashkey-string.patch
- 650-package-version.patch
- 651-package-version-simplify.patch
- 652-package-version-simplify-further.patch
- 653-package-version-warning.patch
- 660-version-stamp.patch
- 689-vc-mtime.patch
- 755-clean-temp-hashkey.patch
- 795-string-desc-rename-functions.patch
- 796-vc-mtime-less-read.patch
- 797-vc-mtime-add-api.patch
- 798-vc-mtime-add-api.patch
- 799-vc-mtime-old-git.patch
- 900-str_startswith-module.patch
- 901-str_endswith-module.patch
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/16522
Signed-off-by: Robert Marko <robimarko@gmail.com>
Move to the June 2025 update of the first 2025 stable branch
in order to better support updating gettext to the latest versions.
More instances of blocking the defining of functions
that use reallocarray() is needed for portability with macOS
as more similar functions have been added.
And one instance is removed as the function was removed.
Refer to a previous commit to explain the need for this.
Manually adjusted patch:
- 150-portable-tdestroy.patch
- 160-flag-reallocarray.patch
- 200-force-disable-after-configure.patch
Removed upstreamed patch:
- 500-acl-function-name.patch
All other patches are automatically refreshed.
Ref: 5ade7ee60 ("tools/gnulib: add macros to skip reallocarray() functions")
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/16522
Signed-off-by: Robert Marko <robimarko@gmail.com>
Snapshot downloads, whether from GNU cgit or GNU gitweb,
are becoming more unreliable and unstable.
Use the GNU gnulib git repository server for downloads
because GNU has disabled cgit server snapshots due to performance issues.
Other possible options include GNU gitweb snapshot or a git bundle download
but these methods either may also have similar performance issues
or require additions to the build system to handle the method
beyond the previous commit.
In case snapshots are to be used again for build tools in the future,
the previous commit reduces the tarball generation attempts
for the original source of snapshots in order to reduce server impact.
A conversation regarding server performance issues and alternatives is linked.
Link: https://lists.gnu.org/archive/html/bug-gnulib/2024-12/msg00124.html
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/16522
Signed-off-by: Robert Marko <robimarko@gmail.com>
The python implementation of gnulib-tool
uses a hidden file as the independent main function.
A copy with glob ('*') does not include hidden files
at the top level directory, so use whole directory copy
instead and remove the patch for a non-hidden python file.
Ensure the directory does not already exist
by attempting uninstall first to remove it
and by not using "install" to create a directory.
Rename the clean target to uninstall, as it handles
the staging directory and not the build directory.
Removed:
- 021-python-main.patch
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/16522
Signed-off-by: Robert Marko <robimarko@gmail.com>
Bump to the next stable branch with the May 2025 update.
Add a patch to compensate for gnulib-tool being further split up
into independent shell and python implementations
by using a non-hidden version of the main.py file.
Add a patch for the python implementation of gnulib-tool
in order to ignore the required version of autoconf in configure.ac
being lower than the required version of autoconf for gnulib
if the version that is being run exceeds the required version for both,
and adjust existing autoconf version shell script patch to new filename.
Backport a patch for a change in function naming convention
for forward compatibility with tool releases after this stable branch.
Added:
- 020-python-version.patch
- 021-python-main.patch
- 500-acl-function-name.patch
Manually Adjusted:
- 010-autoconf-version.patch
- 160-flag-reallocarray.patch
Existing patches are automatically refreshed.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/16522
Signed-off-by: Robert Marko <robimarko@gmail.com>
Add a module to gnulib to support posix_fallocate()
for macOS and other systems that are missing it.
Apple-specific code is sourced from Mozilla,
and the rest from glibc, both licensed under LGPL.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15690
Signed-off-by: Robert Marko <robimarko@gmail.com>
The gnulib-tool script is written to have a fatal error
whenever the minimum required version of autoconf
for the project that gnulib is being imported into
as defined in configure.ac was less than
the minimum required version required by gnulib.
However, none of this matters if the version of autoconf
that we use is newer than both requirements.
Instead, use functions from the bootstrap script
to check for the version of autoconf being used
and print a warning whenever this case occurs.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15690
Signed-off-by: Robert Marko <robimarko@gmail.com>
Overriding variables used in both the macros and the headers
like setting REPLACE_FCNTL to 0 while invoking Make causes the
function aliases like rpl_fcntl() to not be defined,
however the object may still be built with the fnctl() function.
Usually this is enough for building while avoiding
the need to link the resulting libgnu library
to every single other build target for a project
in order to include the gnulib copy of the function,
because in these cases we don't care which version
of the function is used.
However for functions like fcntl() this doesn't work
as it is designed to use either the alias or standard declaration
from gnulib headers in order to be
a wrapper for the native host copy of fcntl()
by containing recursive calls to fcntl() within itself
after undefining the gnulib function declaration.
Overriding the variables used by the header when invoking Make
causes the header's declarations to be blocked,
and this results in the gnulib version of fcntl()
to call itself recursively and indefinitely
leading to segmentation faults.
Fix this by using macros defined with those variables
in order to exclude the function during preprocessing.
While at it, do the same for reallocarray()
so that the configure option --avoid=reallocarray
and Make variable REPLACE_REALLOCARRAY set to 0
would have a similar effect.
In the future this patch can be expanded to include
more functions and some version of this may be upstreamable.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15690
Signed-off-by: Robert Marko <robimarko@gmail.com>
The gnulib fts header is meant to not be overwritten
in any way by the host system's copy of fts.h
and was therefore given a unique name instead.
This is fine if the built libgnu library is directly linked
with the target library, but if we want to keep them isolated
we end up having the definitions being mangled anyway
when the next object to link against included the fts.h header.
On some macOS platforms, the use of __DARWIN_INODE64
is messing with the link name for fts functions, resulting in:
Undefined symbols for architecture x86_64:
"_rpl_fts_close$INODE64", referenced from:
...
Create a local fts header for gnulib
that completely blocks the macOS host fts header.
An alternative and more upstream friendly fix would be
to rename fts_.h to fts.h and add the macOS-only
include guard to that file within it's own include guard,
but that would be a massive patch, so do this for now.
Tested-by: Georgi Valkov <gvalkov@gmail.com> # macOS
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Link: https://github.com/openwrt/openwrt/pull/15368
Signed-off-by: Robert Marko <robimarko@gmail.com>
The tdestroy() function, which is a GNU extension to the standard C
library, is defined in gnulib in tsearch.c but is missing it's
corresponding declaration in search.in.h by being completely missing...
This patch is large but upstreamable, including all of the macros and
conditionals and configure checks that upstream GNU would expect for
portable support, like using the @@ placeholder/substitution method to
determine whether or not to have declarations based on whether or not
tdestroy() is already declared within the standard headers of the default
include paths.
There were also some typedefs and aliases missing, along with the warnings
and preprocessor exceptions that need to be added for consistency with the
usage of the rest of the functions in the files.
Tested-by: Georgi Valkov <gvalkov@gmail.com> # MacOS
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
For modules that depend on the reallocarray module, like ialloc, xalloc,
and safe-alloc, it was not possible to skip importing the reallocarray
module as they all contained at least one function that called
reallocarray() and would cause build failure if the host system didn't
declare it.
This upstreamable patch adds macros that toggle whether to define
functions that depend on reallocarray() based on whether the reallocarray
module is being imported.
Tested-by: Georgi Valkov <gvalkov@gmail.com> # MacOS
Signed-off-by: Michael Pratt <mcpratt@pm.me>
Signed-off-by: Tony Ambardar <itugrok@yahoo.com>
These 7 macro files are provided by gnulib,
so we can now replace these statically stored copies
with the latest copy from GNU sources.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
By having a local copy of gnulib, we can:
import the latest macro fixes into any package,
get rid of some statically stored macros that were otherwise missing,
bootstrap GNU tools with the latest relevant source
without having to wait for a release or rely on git submodules,
and possibly more...
The patch assists in bootstrapping by ignoring
the building of po files using gettext,
and also to allow a user-defined path to a program
to include parameters.
Signed-off-by: Michael Pratt <mcpratt@pm.me>
