I never really looked at the joinpath() function so I just realized it
essentially does os.path.normpath(os.path.join(...)) unlike what it's
doc string says.
I have applied this patch removing it and preferring os.path.join()
which should help make things more clear IMO. The os.path.normpath()
isn
On 2024-06-14 16:51, Bruno Haible wrote:
getpagesize.texi:45: warning: accent command `@,' must not be followed by
new line
Thanks, fixed. I thought I had run 'makeinfo' but evidently I missed the
warning.
On 2024-06-14 15:45, Bruno Haible wrote:
I propose to remove '.cppi-disable'.
Sounds good to me. I don't use cppi.
This patch adds Gnulib documentation for the functions that were added
in glibc 2.35. Many of them are specified in ISO C 23, which is why
I add them under doc/posix-functions/.
2024-06-14 Bruno Haible
doc: Update for glibc 2.35.
* doc/glibc-functions/_dl_find_object.texi: New
> 2024-06-14 Bruno Haible
>
> doc: More updates for glibc 2.34.
And some more. It's unlikely that we will ever have to deal with
these functions. Just for completeness.
2024-06-14 Bruno Haible
doc: More updates for glibc 2.34.
* doc/glibc-functions/ns_name_compress.
In the doc updates for glibc 2.34, I missed the added function 'close_range'.
It exists in
glibc ≥ 2.34, FreeBSD ≥ 13.0, Cygwin ≥ 3.5.0, Android API level ≥ 34.
2024-06-14 Bruno Haible
doc: More updates for glibc 2.34.
* doc/glibc-functions/close_range.texi: New file.
Paul Eggert wrote:
> +On typical platforms this function's API limits page sizes to at most 1 GiB@,
This produces a warning (with makeinfo version 6.8):
getpagesize.texi:45: warning: accent command `@,' must not be followed by new
line
Bruno
Paul Eggert did:
> * lib/.cppi-disable: Remove getpagesize.h.
This file had not been touched for more than 16 years.
When I run
$ cd lib
$ cppi -c *.h *.c
it emits 23812 diagnostics. So, clearly, 'cppi' does not reflect the
indentation habits that we are using for many years.
I propose to re
On 6/14/24 13:46, Collin Funk wrote:
Isn't sysconf not available on Windows?
Yes, it's not. But the sysconf uses in question should be protected by
#ifdef and so not compiled on MS-Windows, so it shouldn't be a problem.
Maybe implementing _SC_PAGESIZE and some other simple ones would work
* lib/xsize.h (xsum): Port to oddball platforms where SIZE_MAX <=
INT_MAX. Although no such platform is a current Gnulib porting
problems, it’s easy enough to port to them.
---
ChangeLog | 5 +
lib/xsize.h | 18 ++
2 files changed, 19 insertions(+), 4 deletions(-)
diff --g
On 6/14/24 1:19 PM, Paul Eggert wrote:
> I noticed when looking into something else that we can simplify
> some getpagesize-using code by removing some dependencies
> and using sysconf (_SC_PAGESIZE) where that might be safer.
> While I was at it, I updated the getpagesize doc and
> fixed what seem
* m4/memchr.m4 (gl_FUNC_MEMCHR):
* m4/strncat.m4 (gl_FUNC_STRNCAT):
Use sysconf (_SC_PAGESIZE) instead of getpagesize, as it’s
in POSIX and there’s no point to worrying about getpagesize
performance advantages here.
---
ChangeLog | 7 +++
m4/memchr.m4 | 4 ++--
m4/strncat.m4 | 4 ++--
3 f
* tests/test-sigsegv-catch-segv1.c, tests/test-sigsegv-catch-segv2.c:
(SIGSEGV_FAULT_ADDRESS_ROUNDOFF_BITS):
Use sysconf (_SC_PAGESIZE) instead of getpagesize, as this code is
compiled only on platforms that have _SC_PAGESIZE and POSIX
prefers the latter.
---
ChangeLog| 11
* tests/test-free.c (main): Use sysconf (_SC_PAGESIZE) instead of
getpagesize, as this code is compiled only on platforms that have
_SC_PAGESIZE and POSIX prefers the latter.
---
ChangeLog | 5 +
tests/test-free.c | 2 +-
2 files changed, 6 insertions(+), 1 deletion(-)
diff --git a/Ch
* modules/memcasecmp-tests, modules/memchr-tests:
* modules/memchr2-tests, modules/memcmp-tests:
* modules/memmem-tests, modules/memrchr-tests:
* modules/rawmemchr-tests, modules/strncat-tests:
* modules/strnlen-tests, modules/strstr-tests:
* modules/unistr/u16-chr-tests, modules/unistr/u16-cmp-tes
* tests/test-ssfmalloc.c (getpagesize) [__hpux]: Remove decl.
(init_pagesize): Use sysconf instead of getpagesize.
---
ChangeLog | 4
tests/test-ssfmalloc.c | 13 ++---
2 files changed, 6 insertions(+), 11 deletions(-)
diff --git a/ChangeLog b/ChangeLog
index 8dcf7133a9
* doc/glibc-functions/getpagesize.texi:
Mention portability problems not fixed by Gnulib.
---
ChangeLog| 6 ++
doc/glibc-functions/getpagesize.texi | 13 +
2 files changed, 19 insertions(+)
diff --git a/ChangeLog b/ChangeLog
index f44aab79f6..9d06cc7ea
I noticed when looking into something else that we can simplify
some getpagesize-using code by removing some dependencies
and using sysconf (_SC_PAGESIZE) where that might be safer.
While I was at it, I updated the getpagesize doc and
fixed what seems to be an obvious porting bug (pointers
do not a
* lib/.cppi-disable: Remove getpagesize.h.
---
ChangeLog | 3 +++
lib/.cppi-disable | 1 -
2 files changed, 3 insertions(+), 1 deletion(-)
diff --git a/ChangeLog b/ChangeLog
index 9d06cc7eac..7b94a3ef80 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,8 @@
2024-06-14 Paul Eggert
+
* m4/getcwd-abort-bug.m4 (gl_FUNC_GETCWD_ABORT_BUG): Don’t invoke
gl_CHECK_FUNC_GETPAGESIZE or #define or use HAVE_GETPAGESIZE.
Protect ‘bug_possible’ with ‘defined _SC_PAGESIZE’ instead.
That’s good enough as the buggy platforms all define _SC_PAGESIZE.
* modules/getcwd (Files): Remove m4/getpages
* m4/mprotect.m4 (gl_FUNC_MPROTECT_WORKS): Don’t assume pointers
fit into unsigned long, as this is not true on CheriBSD. Prefer
sysconf (_SC_PAGESIZE) to getpagesize.
---
ChangeLog | 5 +
m4/mprotect.m4 | 24 ++--
2 files changed, 15 insertions(+), 14 deletions(-)
I'm still working on allowing *printf results of size > 2 GiB.
One of the issues here is what to do when a width > INT_MAX
is specified directly in the format string, for example
"%030d".
ISO C § 7.23.6.1 only says that this should be a "nonnegative decimal integer",
but doesn't give a li
Hi Collin,
> The following file is mentioned in "Information for maintainers of GNU
> software":
>
> announce-gen [1]
>
> The following files are mentioned "GNU coding standards" [2]:
>
> gitlog-to-changelog [2]
> vcs-to-changelog.py [2]
>
> It reads like they are maintained by Gnu
I wrote:
> config.libpath --help -> Bruno
> config.rpath --help-> Bruno
> declared.sh --help -> Bruno
> install-reloc --help -> Bruno
> libtool-next-version --help-> Bruno
> libtool-reloc --help -> Bruno
> reloc-ldflags --help
24 matches
Mail list logo
