Hi,
This change:
* df now displays a mount point (usually `/') for non-mounted
character-special and block files
is counter-intuitive, because it produces
$ df /dev/hda
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/hda 2208864 2127604 81260
it is no problem because gnulib doesn't compile regex with
MBS_SUPPORT.
- In strftime.c it is no problem gnulib doesn't compile strftime with
COMPILE_WIDE.
Bruno
2003-08-11 Bruno Haible [EMAIL PROTECTED]
* vasnprintf.c (local_wcslen): New function, for Solaris 2.5.1.
(vasnprintf
Bob Proulx wrote:
But sed and tr and other utilities just use the locale data provided
on the system by glibc among other places. These programs are table
driven by tables that are not part of these programs. This is why
locale problems are global problems across the entire system of
Jim Meyering wrote:
http://www.loria.fr/~zimmerma/records/ecmnet.html
http://www.loria.fr/~zimmerma/records/ecm-5.0.3.html
I like that idea. Thanks!
I've always wanted factor to accept arbitrarily large inputs.
Although you can input an arbitrarily large integer, the limit of what
the
Hi, Alexander Patrakov,
I've forwarded your bug report to the russian translation team (the email
address indicated in the Language-Team field of the PO file's header). They
are responsible for it.
This message is cross-posted to [EMAIL PROTECTED] because
msgfmt --check-format does not catch
/gethostname.c to add Woe32 support,
- a new src/hostname.c,
- some autoconf macros that you can add to configure.ac or where it fits.
If that's OK, I can also go out and write the documentation update.
2004-01-18 Bruno Haible [EMAIL PROTECTED]
* gethostname.c: Add support for Windows
Jim Meyering wrote:
I've begun looking through these and like what I've seen.
However, I'd prefer to avoid providing short-named options
for --help and --version.
As you like. I did this for compatibility with the Linux /bin/hostname.
I don't know how important these short options are in
Jim Meyering wrote:
But OTOH, I didn't follow compatibility with Linux /bin/hostname in two
points:
- When the machine has multiple long names (i.e. some aliases),
my hostname -f prints them all, one per line. Linux /bin/hostname -f
prints only the first one, but has an extra
Paul Jarc wrote:
When it comes to DNS, there can be multiple names, but there need not
be any primary name among them. There can be, and often is, but
it's also possible that all may have equal standing.
Then why is does gethostbyname() return one of the names as h-h_name,
and the others as
Paul Eggert wrote:
I installed this obvious little patch in both gnulib and coreutils:
2004-09-08 Paul Eggert [EMAIL PROTECTED]
* c-strtod.c (C_STRTOD) [!defined LC_ALL_MASK]: Set LC_ALL to C,
not just LC_NUMERIC, to avoid the unlikely possibility of mixed
locales
Paul Eggert wrote:
wouldn't it be better simply to
remove the linebuffer module and rewrite all its invokers to use
getline? I don't see the point of the linebuffer module
The linebuffer module is an object-oriented line buffering code;
getline serves a similar purpose but is written in
Jim Meyering wrote:
Do any of you know of a system
(reasonable porting target) for which errno.h does not define EBADF?
No. Never had a portability problem like this. EBADF is so basic that even
the most primitive Unix emulation has it (including mingw).
Bruno
Paul Eggert wrote:
A couple of other things. The Bison documentation says that
YYSTACK_USE_ALLOCA must be defined to 0 if it is defined to anything.
But then the second to fourth line of the following bison output are useless.
# ifdef YYSTACK_USE_ALLOCA
# if YYSTACK_USE_ALLOCA
# define
Eric Blake wrote:
Actually, the tarball includes AM_GNU_GETTEXT_VERSION(0.13.1), the latest
version of autopoint is 0.14.1, but some of the gettext .m4 files are
newer than the 0.14.1 version. To get a good reconf, I had to use
`AUTOPOINT=true autoreconf -fiv', skipping autopoint, because
Jim Meyering wrote:
Thanks for the patch, but that `cure' seems worse than the disease.
I agree. The real good way to make this work on POSIX systems (without
assuming multithreading) is to use fork().
- Let the parent process fork a child process, block the signals SIGHUP,
SIGQUIT,
James Youngman wrote:
Paul +static int
Paul +fts_compar (void const *a, void const *b)
Paul +{
Paul/* ... */
Paul + return pa[0]-fts_fts-fts_compar (pa, pb);
Paul +}
... compilers are likely to be able to inline the actual
subroutine call away in any case.
How should this be
Paul Eggert wrote:
! # define verify_expr(R) ((void) sizeof (verify_type__ (R)))
This doesn't work in C++: gcc gives an error
error: ISO C++ forbids defining types within sizeof
However, Jim's first version with the NULL pointer works in C++ too:
#define verify_expr(R) (void) ((verify_type__
Hi,
Here's a case where GNU df 5.2.1 displays figures that are totally off
the reality.
[EMAIL PROTECTED]:/smb/ibook/data/temp $ df --version
df (coreutils) 5.2.1
...
[EMAIL PROTECTED]:/smb/ibook/data/temp $ df .
Filesystem 1K-blocks Used Available Use% Mounted on
//ibook/bruno
Paul Eggert wrote:
I installed this
patch into CVS coreutils. I don't think it'll fix the problem in
general, but does it fix your particular problem?
Thanks. The patch does have the effect that, in my case,
[EMAIL PROTECTED]:/smb/ibook/data/temp $ df .
calls
statfs(.,
to Yoann's fix.
Bruno
2005-09-26 Bruno Haible [EMAIL PROTECTED]
* modules/mbchar (Include): Mention that HAVE_WCHAR_H HAVE_WCTYPE_H
is necessary.
(lib_SOURCES): Remove mbchar.c.
* modules/mbfile (Include): Mention that HAVE_MBRTOWC is necessary.
(Files): Add m4
Eric Blake wrote:
Perhaps GNU extensions PRIoSIZE, PRIuSIZE, PRIxSIZE, PRIXSIZE could be
added to GNU gettext. That would mean we could write this instead:
printf (_(The size is %PRIuSIZE.\n), size);
which would be nicer than either of the above.
But your idea of formalizing
Paul Eggert wrote:
(__strndup): Revert to KR-style function dfns, the glibc style.
Huh? We know the problems of KR-style function definitions: arguments
of type 'float', 'short' and 'char' are implicitly promoted, leading to
a clash with the function prototype. Empty argument lists allow
$gl_cv_func_strndup = no; then
Was this intended redundancy or a typo?
Bruno
2006-05-30 Ralf Wildenhues [EMAIL PROTECTED]
Bruno Haible [EMAIL PROTECTED]
* strndup.m4 (gl_FUNC_STRNDUP): Replace the AC_REPLACE_FUNCS with a
check for the declaration of strnlen
Ralf Wildenhues wrote:
Tested on AIX 4.3.3, 5.1, 5.2 (first has strnlen and strndup broken,
second has strnlen fixed, last has both fixed)
Hmm? My results for AIX 5.1 differ: On AIX 5.1.0.0 I get
checking whether strndup is declared... yes
checking for working strndup... no
checking whether
Ralf Wildenhues wrote:
Hmm? My results for AIX 5.1 differ: On AIX 5.1.0.0 I get
checking whether strndup is declared... yes
checking for working strndup... no
checking whether strnlen is declared... yes
checking for working strnlen... no
(lslpp -L). The one I tested on has:
Paul Eggert wrote:
This burned another hour of my time. I suppose I should bump the
priority of getting bootstrapping to work with coreutils.
The bidirectional merge between gettext and gnulib after gettext-0.15
took me 5 hours. Then I switched to using gnulib-tool. The number of
files
On Linux/x86 (Linux 2.4.21, glibc-2.3.6) I get:
make[3]: Entering directory `/build/coreutils-6.0/tests/dd'
dd iflag=noatime updated atime; O_NOATIME bug in your kernel?
FAIL: misc
Your announcement says that O_NOATIME is only supported in Linux = 2.6.8.
So this should be an XFAIL, not a FAIL,
--help output to stderr.
The appended patch fixes the problem.
Bruno
2006-08-16 Bruno Haible [EMAIL PROTECTED]
* Makefile.maint (gzip_rsyncable): Throw away stderr output of
gzip --help.
*** Makefile.maint 15 Aug 2006 12:13:04 - 1.243
--- Makefile.maint 16
Hi,
The c99-to-c89.diff mentioned in the announcement is not enough to
build with Solaris cc on Solaris 7. I get an error
source='fts.c' object='fts.o' libtool=no \
DEPDIR=.deps depmode=none /bin/bash ../build-aux/depcomp \
cc -O -DHAVE_CONFIG_H
Hi,
On Solaris 7, building with cc -O, I get two test failures:
chmod/setgid fails
Here are details:
$ VERBOSE=yes gmake check TESTS=setgid
gmake check-TESTS
gmake[1]: Entering directory `/opt/coreutils-6.0/tests/chmod'
Paul Eggert wrote:
I think this is the right sort of fix, though I'm puzzled as to why
the code set LC_ALL to the empty string in the first place.
...
Do you recall why they were set to '' back then?
Yes, my archives tell me: On 2000-10-15, I sent Jim a mail, reporting
testsuite failures,
Egmont Koblinger asked:
What is the correct approach if the number isn't presented in the string, as
in his This item vs. These items case? Should ngettext be used in this
case? Or this piece of code a good programming approach:
if (n == 1) printf(_(This item...)); else printf(_(These
Paul Eggert wrote:
If the intent is that such usages are OK, how
about if we reduce the confusion by documenting the intent?
Yes, I added the same example, at a different place and without
talking too much about compatible format strings and conversion
specifications. These technical terms are
check. After it, you might
want to rename MOUNTED_GETMNTINFO to MOUNTED_GETMNTINFO1.
2006-08-18 Bruno Haible [EMAIL PROTECTED]
Add support for NetBSD 3.0.
* m4/ls-mntd-fs.m4 (gl_LIST_MOUNTED_FILE_SYSTEMS): Also check for
sys/statvfs.h. When getmntinfo was found, check
, NetBSD 3.0 has a 'struct statvfs' with member
f_fstypename and no f_basetype, and 'struct statfs' doesn't exist any more.
This patch works on NetBSD 3.0, and passes make check.
2006-08-18 Bruno Haible [EMAIL PROTECTED]
Add support for NetBSD 3.0.
* m4/stat-prog.m4
Hi,
I'm using shred-4.5.7 to clean disk partitions. Often a partition has an odd
number of 512-byte blocks; Linux fdisk indicates such partitions with a '+'.
Shred, when used on such a partition, gives three error messages:
attempt to access beyond end of device
03:06: rw=0, want=4200968,
: automake
drops the distdir rule if you make it depend on something).
2006-08-19 Bruno Haible [EMAIL PROTECTED]
Allow make check when some programs are not built.
* src/Makefile.am (programs): New variable.
(check-AUTHORS): Depend on $(programs), not $(all_programs
fchmod. This fixes it.
2006-08-19 Bruno Haible [EMAIL PROTECTED]
BeOS portability.
* lib/dirchownmod.c (dirchownmod): Fall back to lchmod or chmod if
fchmod doesn't exist.
*** lib/dirchownmod.c.bak 2006-07-17 05:05:23.0 +0200
--- lib/dirchownmod.c 2006-08-19
Paul Eggert wrote:
Also, we need to define HAVE_FCHMOD when fchmod is available.
Yes, sorry. Coreutils already provides the HAVE_FCHMOD in config.h.in;
I had forgotten that dirchownmod is also in gnulib.
Bruno
___
Bug-coreutils mailing list
On BeOS, coreutils wouldn't install the 'uptime' program. But BeOS has a
system call that returns the boot time. This makes it possible to port
the 'uptime' program.
2006-08-19 Bruno Haible [EMAIL PROTECTED]
BeOS portability.
* m4/boottime.m4 (GNULIB_BOOT_TIME): Test also
) and readutmp.c.
2006-08-19 Bruno Haible [EMAIL PROTECTED]
* m4/readutmp.m4 (gl_READUTMP): Compile readutmp.c only if utmp.h or
utmpx.h exists.
* lib/readutmp.h: Skip most definitions if neither utmp.h nor
utmpx.h exists.
--- m4/readutmp.m4.bak 2005-03-30 09:41
, but I don't know what value could be used for
PF_UNSPEC if we were to make a sys/socket.h override.
2006-08-19 Bruno Haible [EMAIL PROTECTED]
BeOS portability.
* lib/getaddrinfo.c (PF_INET): Fallback define.
(validate_family): Don't recognize PF_UNSPEC if it doesn't
Egmont Koblinger wrote:
+ @smallexample
+ printf (ngettext (One file removed, %d files removed, n), n);
+ @end smallexample
Please don't do so, please don't mention this in the manual!...
- One file, 2 files, 3 files is really inconsistent and really ugly.
Users view something as
in.)
make[3]: *** [ln.o] Error 1
make[3]: Leaving directory `/boot/home/gnubuild/coreutils-6.0/src'
BeOS has no quotas and no EDQUOT. Fortunately it is easy to fix.
2006-08-19 Bruno Haible [EMAIL PROTECTED]
BeOS portability.
* src/ln.c (EDQUOT): Fallback define.
--- src/ln.c.bak
in.)
BeOS has a sigaction function with SA_NOCLDSTOP, but no SA_RESTART. This
workaround does it.
2006-08-19 Bruno Haible [EMAIL PROTECTED]
BeOS portability.
* src/ls.c (SA_RESTART): Fallback define.
--- src/ls.c.bak2006-08-19 13:18:35.0 +0200
+++ src/ls.c2006
uses, and to transform the lib/fsusage.c code
into a gnu_statvfs function that could be used by both df and stat.
Does this seem worthwhile? Can the SVR2 code (this is the only code in
lib/fsusage.c that makes use of the 'disk' argument) be dropped in the process?
2006-08-19 Bruno Haible [EMAIL
Jim Meyering asks:
I'm curious: what's your motivation for using it?
Portability testing. Through the BeOS porting, I found the following
problems not limited to BeOS:
- check-AUTHORS: applies to any platform that doesn't build all of the
programs.
- mbchar.h problem: applies to any
bytes.
But we cannot do anything about it.
Perhaps you can use 'strace' to check my guesses above?
Too late. The machine's already wiped out, sorry.
The appended patch might fix these cases and is at the same time safe.
2006-08-24 Bruno Haible [EMAIL PROTECTED]
* src/shred.c (dopass
Paul Eggert wrote:
what is the
recommended way to use AM_GNU_GETTEXT([external], ...), gnulib-tool,
and autopoint/autoreconf ... autopoint brings in
yet another (incompatible) copy of them.
Currently, there is a general problem: After a release is made, if users
do a gnulib-tool --update,
+ dd iflag=nofollow if=dd-sym.20477 count=0
+ fail=1
...
It seems that the nofollow test doesn't work?
What does ktrace tell you about the system calls that were executed?
I assume MacOS X has ktrace?
cd src
ln -s dd.c sym
ktrace ./dd iflag=nofollow if=sym count=0
kdump
, and only the relevant
flags asserted. If family isn't asserted, it usually means take any
family.
I see. Then what about this patch? It compiles fine on BeOS.
2006-08-26 Bruno Haible [EMAIL PROTECTED]
Simon Josefsson [EMAIL PROTECTED]
* getaddrinfo.c (PF_INET, PF_UNSPEC
Paul Eggert wrote:
2006-08-23 Paul Eggert [EMAIL PROTECTED]
* src/stat.c (HAVE_STRUCT_STATXFS_F_FSID___VAL): Define. This
macro was being used without being defined.
Compiling the current coreutils CVS on MacOS X 10.3.9 now gives this error:
stat.c: In function
Paul Eggert wrote:
what is the
recommended way to use AM_GNU_GETTEXT([external], ...), gnulib-tool,
and autopoint/autoreconf in such a way that you don't get the
following files into your m4 directory afterwards?
glibc2.m4 intdiv0.m4 inttypes-h.m4 inttypes-pri.m4 lcmessage.m4
lock.m4
', because the purpose of these subfields is just to split
8 bytes of data into two integers. So I propose this patch.
With this patch, thanks to your fixes regarding O_NOFOLLOW and isapipe(),
make check succeeds!
2006-08-31 Bruno Haible [EMAIL PROTECTED]
* src/stat.c (print_statfs): Use
yield numbers 3, 4, 5, ...)
2006-08-20 Bruno Haible [EMAIL PROTECTED]
* src/remove.c (remove_cwd_entries): Disable assertion on BeOS.
--- coreutils-6.0-beos/src/remove.c 2006-08-20 16:00:08.0 +0200
+++ coreutils-6.0-beos/src/remove.c 2006-08-20 18:08:58.0 +0200
Andreas Schwab wrote:
--- 404,422
case 'i':
{
! uintmax_t fsid;
! /* On many BSD systems, fsd.f_fsid is of type fsid_t which is a
! struct type:
! struct { int32_t val[2]; }
! or struct { int32_t __val[2]; }
! or struct { int32_t
-19 Bruno Haible [EMAIL PROTECTED]
* lib/fts.c (fts_open): If fchdir() does not exist, set FTS_NOCHDIR
and unset FTS_CWDFD.
(fts_stat): If fchdir() does not exist, use lstat instead of fstatat.
* src/remove.c (AD_pop_and_chdir): If fchdir() does not exist, try
Paul Eggert wrote:
On platforms lacking fchdir,
put a wrapper around 'open' so that we can keep track of which file
descriptors correspond to directories. The 'open' wrapper puts the
name of the opened directory into a hash table. (The name must be
absolute, so 'open' may need to do the
Eric Blake wrote:
It is not possible to know what macros will be invoked later in the file.
But with diversions, it is possible for macros invoked later to emit
shell code that will be invoked earlier, by making sure that
gt__NEED_NGETTEXT_FU expands to shell code in a lower diversion number
Gary V. Vaughan wrote:
Untested:
m4_define([gt__NEED_NGETTEXT_FU],
[shell code to execute after expansion of AM_GNU_GETTEXT,
whenever AM_GETTEXT_NEED_NGETTEXT is present somewhere])
AC_DEFUN([AM_GNU_GETTEXT],
[
...current contents of AM_GNU_GETTEXT definition...
dnl make sure
Hi,
Jim Meyering wrote:
+static inline bool
+has_trailing_slash (char const *file, size_t len)
+{
+ /* Don't count / as having a trailing slash. */
+ if (len = 1)
+return false;
+
+ char last = file[len - 1];
+ return ISSLASH (last);
Since you use ISSLASH, you apparently care
Paul Eggert wrote:
It's a bit weird, though, that Tru64 has getnameinfo declared but
getaddrinfo is not defined. Is that correct, or am I missing
something?
OSF/1 4.0 declares neither getnameinfo nor getaddrinfo.
OSF/1 5.1 declares both in netdb.h but has this:
#if defined (_SOCKADDR_LEN)
Jim Meyering wrote:
The problem that causes the coreutils build failure is due in part
to the way the bootstrap script invokes gnulib-tool, then moves/links
the resulting files into our canonical lib/ and m4/ directories.
Since coreutils is perhaps unusual in doing things this way,
... *
Jim Meyering wrote:
I'll use 180.
The lower we go, the more of a performance penalty
we impose for directories with very many entries.
I tried the value 180. It worked fine in some cases, but still failed in
others:
$ tar xf /Volumes/ExtData/bin.x86-linux/cross/cross-hppa.tar.gz
$ ll
Jim,
Is there any type of file system where readdir works?
I tried only HFS+ mounts (on two different volumes).
What version of Darwin are you using?
Darwin 7.9.0 = MacOS X 10.3.9.
I see no failure with Darwin-8.7.0, so I might
add a run-test (like coreutils' old readdir.m4),
if it's
Hello Jim,
Is there any type of file system where readdir works?
Yes. It does work on vfat file systems. No readdir bug reproducible there.
What version of Darwin are you using?
Darwin 7.9.0 = MacOS X 10.3.9.
I wrote:
In the run test, you can use /bin/rm, since /bin/rm has the same bug
as
Hi Jim,
The Darwin HFS+ bug is even reproducible on Linux, on NFS mounts from a
Darwin 10.3.9 machine. Here is that same directory, of which Darwin's
readdir() bug occurred after 178 removals. Here it occurs already after
13 removals on average:
$ ll charmaps | wc -l
195
$ rm -r charmaps
rm:
Hi Jim,
You mean you've found *another* problem with GNU rm?
Is so, please provide details.
The second bug is harder to reproduce.
I have a big tar file with hard links in it.
$ tar tvf cross-hppa.tar | grep ' link '
hrwxr-xr-x bruno/user0 2002-06-02 01:20:05
In fact, I am unable to measure any significant performance difference.
Nice! It'd have been a pain if a bug on one system forced a significant
performance hit on all other systems.
Bruno
___
Bug-coreutils mailing list
Bug-coreutils@gnu.org
On Darwin-7.9.0 with CPPFLAGS=-Wall: Builds fine. All tests pass. The
rm -rf HFS+ bug workaround works fine. The rm -rf hardlinks / NFS bug
is still open.
The following warnings were seen:
In file included from openat.c:285:
at-func.c: In function `fchownat':
at-func.c:39: warning: implicit
Paul Eggert wrote:
However, I can't test this easily since I don't have such a kernel.
...
It's not an ideal fix, but it's the best I could
think of offhand.
Thanks! coreutils + gnulib from Saturday now pass make check entirely
fine on the Linux-2.4.21 machine.
Bruno
gl_AC_TYPE_LONG_LONG.
This is unfortunately two patches in one, with few comments.
I'm applying the first part first:
2006-10-11 Paul Eggert [EMAIL PROTECTED]
Bruno Haible [EMAIL PROTECTED]
* lib/allocsa.h (sa_alignment_longlong, sa_alignment_max): Test
HAVE_LONG_LONG_INT instead
Hi,
Compiling coreutils-6.4 with gcc-4.1 and
./configure --prefix=/packages/gnu \
CPPFLAGS=-Wall -Wformat=2 -Wmissing-field-initializers
-Wmissing-format-attribute -Wpointer-arith -Wstrict-aliasing=2 -Wwrite-strings
\
CFLAGS=-O2 -g -Wbad-function-cast -Wdeclaration-after-statement
yields
a warning:
fts.c:1076: warning: ISO C90 forbids mixed declarations and code
gnulib still assumes C89 only. Here is a fix.
Jim, OK to apply?
2006-10-23 Bruno Haible [EMAIL PROTECTED]
* fts.c (fts_build): Move variable declaration, for C89 compliance.
*** fts.c.bak 2006-10-13 15:41
protect yourself against short-lived bugs.
2006-11-20 Bruno Haible [EMAIL PROTECTED]
* gettext.m4 (AM_GNU_GETTEXT): Revert 2005-07-28 patch: Use
changequote instead of pairs of brackets.
Reported by Andreas Schwab [EMAIL PROTECTED].
diff -r -c3 --exclude='*.po*' --exclude
FYI: This bug is now fixed not only in gnulib, but also in gettext-0.16.1.
2006-11-20 Bruno Haible [EMAIL PROTECTED]
* gettext.m4 (AM_GNU_GETTEXT): Revert 2005-07-28 patch: Use
changequote instead of pairs of brackets.
Reported by Andreas Schwab [EMAIL PROTECTED
but with ISO-8859-1
characters in his filenames cares about it...
Bruno
*** lib/wctype_.h 22 Dec 2006 00:21:54 - 1.1
--- lib/wctype_.h 22 Dec 2006 16:19:33 -
***
*** 18,30
/* Written by Bruno Haible and Paul Eggert. */
! /* iswctype, towctrans
Paul Eggert wrote:
2006-12-21 Paul Eggert [EMAIL PROTECTED]
* MODULES.html.sh: New module wctype.
* lib/wctype_.h, m4/wctype.m4, modules/wctype: New files.
* lib/fnmatch.c: Don't bother to include wchar.h before
wctype.h, since the new wctype module should fix
Hi,
When 'sort' is fed input without spaces, one would expect that
sort -t . [OTHER-OPTIONS]
produces the same sort order as
tr '.' ' ' | sort [OTHER-OPTIONS]
Right? That's not the case with GNU sort 6.7:
$ printf '8.0.2\n8.1.0\n8.0.3\n10.1.0\n8.0.11\n11.0.0\n' input
$ cat input |
it, and makes all tests pass:
2006-12-26 Bruno Haible [EMAIL PROTECTED]
* src/copy.c (copy_internal): Use mkfifo as a fallback if mknod fails.
Needed on MacOS X.
*** src/copy.c.bak Thu Dec 7 08:01:16 2006
--- src/copy.c Tue Dec 26 11:24:41 2006
***
*** 1686,1697
Hi,
On MacOS X 10.3.9, in a German locale, 3 tests fail because they compare a
localized error message with an expected result in English. Here is a fix.
2006-12-28 Bruno Haible [EMAIL PROTECTED]
* tests/chown/deref: Source lang-default.
* tests/misc/split-a: Likewise
Hi,
When I use the fchdir emulation on a Linux 2.4.x system (that doesn't have
openat() and similar), the coreutils-6.7 tests install/basic-1 and mkdir/p-3
fail. The reason is this part:
$ mkdir -p sub1/d
$ cd sub1/d
$ chmod a-rx ..
$ chmod a-r .
$ ginstall -d rel/a rel/b
ginstall: cannot create
to test this with coreutils, you need to add fchdir to the
module list, use --avoid=canonicalize-lgpl, and drop the fchdir-stub.c.
2006-12-30 Bruno Haible [EMAIL PROTECTED]
* modules/fchdir: New file.
* modules/unistd (Files): Add lib/unistd_.h.
(Makefile.am): Generate
on 'canonicalize-lgpl', which is the small brother of the
'canonicalize' module. coreutils doesn't need to include
lib/canonicalize-lgpl.c and m4/canonicalize-lgpl.m4.
coreutils builds fine with the appended patch.
Bruno
2007-01-14 Bruno Haible [EMAIL PROTECTED]
* bootstrap.conf
The revised gettimeofday module requires IMO the following changes in
gnulib. Opinions? Paul?
--- lib/gettime.c 13 Sep 2006 22:38:14 - 1.7
+++ lib/gettime.c 17 Jan 2007 11:48:37 -
@@ -1,6 +1,6 @@
/* gettime -- get the system clock
- Copyright (C) 2002, 2004, 2005,
Eric Blake wrote:
coreutils does not handle multi-byte locales well.
True,
The problem is that no one has yet written a patch that makes it
easy to handle multibyte locales without penalizing single-byte locales.
There are patches for multibyte locale support for many of the text
utilities,
Vincent Lefevre wrote:
Therefore: can you also show wrong behaviour when you set
LC_ALL=en_US.UTF-8 ?
Yes:
prunille:~/blah export LC_ALL=en_US.UTF-8
prunille:~/blah locale
LANG=POSIX
LC_COLLATE=en_US.UTF-8
LC_CTYPE=en_US.UTF-8
LC_MESSAGES=en_US.UTF-8
LC_MONETARY=en_US.UTF-8
platform (Linux,
Solaris, etc.), the fix needs to be platform independent. Here is such a fix:
2007-01-18 Bruno Haible [EMAIL PROTECTED]
Avoid problems with tabs after non-ASCII characters in some terminals.
* src/ls.c (nonascii_in_this_line): New variable.
(quote_name
Jim Meyering wrote:
As I understand the goal, you'd like to make ls act differently
(outputting spaces, not TABs, for column alignment) on all systems
for each line containing a non-ASCII byte.
Yes, this is what the proposed patch does.
That change would contradict the documentation of -T
Jim Meyering wrote:
Um... it *is* possible to use TABs after non-ASCII bytes and get correct
alignment. The only requirement is that you be using a reasonable
(non-buggy) terminal emulator.
Yes, sure. I was only pointing out that the proposed change wouldn't need
a doc change, because the
Paul Eggert wrote:
Long ago I regularly used terminal emulators that mishandled tabs.
Eventually they got fixed (or I stopped using them).
Long ago I used terminals where the tab stops were customizable, and the
previous user had set them to weird values. At that time, I stopped using
tabs.
Hi Jim,
Joel E. Denny just discovered a bug [1][2] in 'tr' of coreutils-5.2.1 which is
already fixed in coreutils-6.4 or newer. But nevertheless, this indicates
that the test coverage of 'tr' can be improved. Can you add this as a test?
The command that leads to the bug is:
echo
PS: Here are two further test cases.
Command: echo 'abc=' | tr 'a=b' 'A\055B'
Expected result: ABc-
coreutils-5.2.1 result: ABcB
Command: echo 'abc-' | tr 'a\055b' 'A=B'
Expected result: ABc=
coreutils-5.2.1 result: A=c-
Hi Jim,
That's already covered by this test, from tests/tr/Tests.pm:
['bs-055', q|'a\055b' def|, a\055b, 'def', 0],
Ah, cool!
FYI, that bug was fixed in July of 2004. The fix appeared in
coreutils-5.3.0, released over two years ago. People really shouldn't
be using such an old
is 'nfsv3'. The NFS server in this
case is dedicated hardware designed for use as a file server. This
workaround workd for me (causes the test to be skipped).
2007-02-23 Bruno Haible [EMAIL PROTECTED]
* tests/du/slink: Skip the test if executing on an nfsv3 file system.
--- coreutils-6.7
Jim Meyering wrote:
1) Make: bad lock name errors when 'make' enters the src directory.
This was already reported in
http://lists.gnu.org/archive/html/bug-coreutils/2004-02/msg00062.html
I think using bash is one way to work around that.
The problem is with the program named `['.
I used
Eric Blake wrote:
Actually, the code is already correct, it is
just missing the TRANSLATORS comment. For example, in GNU Hello, the code
looks like this:
/* TRANSLATORS: --help output 5 (end)
TRANSLATORS: the placeholder indicates the bug-reporting address
for this
Hi,
Built coreutils-6.9 on AIX 4.3.2 and using these binaries on AIX 5.1.
'ls' and 'stat' show modification times with negative nanosecond values,
on a filesystem of type 'jfs2' (e.g. /tmp).
$ stat abc9
File: `abc9'
Size: 0 Blocks: 0 IO Block: 4096 regular empty file
Jim Meyering wrote:
Well, at least it is not IA64-specific.
To trigger it you need a long double type longer than 8 bytes.
More precisely, it occurs for floating-point data formats in which the
most significant mantissa bit is not hidden.
- It does not occur with IEEE 754 'float', 'double'.
to do this instead.
This should fix it for gnulib-tool. (Untested.)
2007-08-26 Bruno Haible [EMAIL PROTECTED]
* gnulib-tool (func_import): When deciding which files to remove,
consider also dangling symbolic links.
Reported by Eric Blake.
--- gnulib-tool 26 Aug 2007 10:31
1 - 100 of 375 matches
Mail list logo
