Bill Brelsford wrote:
Package: coreutils
Version: 7.5-3
Severity: normal
tail -f no longer works with stdin. E.g. commands such as
somecommand | tail -f -
somecommand | tail -f
tail -f /var/log/kern
fail with the message:
tail: cannot watch `-': No such file
Hi,
Jim Meyering wrote:
Pádraig Brady wrote:
Ondřej Vašík wrote:
As reported in
http://lists.gnu.org/archive/html/bug-coreutils/2009-08/msg00342.html by
Ernest N. Mamikonyan, cp/mv fails to preserve extended attributes for
read-only source files.
Following patch fixes the issue for
The conversion of everything to long doubles internally makes seq a lot
slower than it needs to be in integer cases, I assume from the use of
floating-point multiplication for every line of output:
seq.c:257 x = first + i * step;
$ time seq 100 /dev/null
real0m1.616s
user
Philip Rowlands wrote:
The conversion of everything to long doubles internally makes seq a lot
slower than it needs to be in integer cases, I assume from the use of
floating-point multiplication for every line of output:
seq.c:257 x = first + i * step;
$ time seq 100
Ondřej Vašík wrote:
Ah, I knew I forgot to do something :). Thanks for spotting this.
Restoring to dest_mode ~omitted_permissions done in attached patch,
dropped redirections from the test as well. Additionally - I modified
the copy.c patch a bit - failure of mode change now doesn't mean
There have been disproportionately many bug fixes since coreutils-7.5.
It's an interesting mix of fixes for recent regressions and for a few older
bugs.
coreutils snapshot:
http://meyering.net/cu/coreutils-ss.tar.gz
http://meyering.net/cu/coreutils-ss.tar.xz
Ondřej Vašík wrote:
Ah, I knew I forgot to do something :). Thanks for spotting this.
Restoring to dest_mode ~omitted_permissions done in attached patch,
dropped redirections from the test as well. Additionally - I modified
the copy.c patch a bit - failure of mode change now doesn't mean
Pádraig Brady wrote:
Ondřej Vašík wrote:
Ah, I knew I forgot to do something :). Thanks for spotting this.
Restoring to dest_mode ~omitted_permissions done in attached patch,
dropped redirections from the test as well. Additionally - I modified
the copy.c patch a bit - failure of mode
Hi Jim,
what do you think about the following solution? It avoids to revert to
the old polling mechanism using /dev/stdin instead of - to
inotify_add_watch.
Cheers,
Giuseppe
diff --git a/src/tail.c b/src/tail.c
index e3b9529..016b712 100644
--- a/src/tail.c
+++ b/src/tail.c
@@ -1152,6
Giuseppe Scrivano wrote:
what do you think about the following solution? It avoids to revert to
the old polling mechanism using /dev/stdin instead of - to
inotify_add_watch.
I considered that and discussed the
trade-off in the comment I committed.
There have been systems and configurations
Eric Blake wrote:
According to Jim Meyering on 9/7/2009 3:09 AM:
ls -i now prints consistent inode numbers also for mount points.
This makes ls -i DIR less efficient on systems with dysfunctional readdir,
because ls must stat every file in order to obtain a guaranteed-valid
inode
Ondřej Vašík wrote:
Pádraig Brady wrote:
To minimize side affects perhaps we should only do the chmod(600)
if (geteuid () != 0 !access (src_name, W_OK)) ?
Good idea, it would reduce possibility of security leak, playing with
access rights is always a bit dangerous (although here we play
Eric Blake e...@byu.net writes:
In other words, are we guaranteed that mount points can only occur
atop directories, and that we can avoid the stat() for regular files?
You can also (bind-)mount regular files.
Andreas.
--
Andreas Schwab, sch...@linux-m68k.org
GPG Key fingerprint = 58CA 54C7
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Andreas Schwab on 9/7/2009 5:17 AM:
Eric Blake e...@byu.net writes:
In other words, are we guaranteed that mount points can only occur
atop directories, and that we can avoid the stat() for regular files?
You can also (bind-)mount
Jim Meyering wrote:
Eric Blake wrote:
According to Jim Meyering on 9/7/2009 3:09 AM:
ls -i now prints consistent inode numbers also for mount points.
This makes ls -i DIR less efficient on systems with dysfunctional readdir,
because ls must stat every file in order to obtain a
Pádraig Brady wrote:
Ondřej Vašík wrote:
Pádraig Brady wrote:
To minimize side affects perhaps we should only do the chmod(600)
if (geteuid () != 0 !access (src_name, W_OK)) ?
Good idea, it would reduce possibility of security leak, playing with
access rights is always a bit
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Michal Svoboda on 9/6/2009 5:33 AM:
When doing cp -va I can see neat quotes (depending on locale), as in
„blah“, but the arrow is still composed of a dash and a greater-than
symbol, as in -. Is there any plan to make the arrow also
Hi,
On Mon, Sep 07, 2009 at 07:23:12AM -0600, Eric Blake wrote:
According to Michal Svoboda on 9/6/2009 5:33 AM:
When doing cp -va I can see neat quotes (depending on locale), as in
„blah“, but the arrow is still composed of a dash and a greater-than
symbol, as in -. Is there any plan to
Ondřej Vašík wrote:
Pádraig Brady wrote:
Ondřej Vašík wrote:
Pádraig Brady wrote:
To minimize side affects perhaps we should only do the chmod(600)
if (geteuid () != 0 !access (src_name, W_OK)) ?
Good idea, it would reduce possibility of security leak, playing with
access rights is always
Erik Auerswald wrote:
Hi,
On Mon, Sep 07, 2009 at 07:23:12AM -0600, Eric Blake wrote:
According to Michal Svoboda on 9/6/2009 5:33 AM:
When doing cp -va I can see neat quotes (depending on locale), as in
„blah“, but the arrow is still composed of a dash and a greater-than
symbol, as in -.
On Mon, 7 Sep 2009, Eric Blake wrote:
This was discussed last month. The verdict is no.
http://lists.gnu.org/archive/html/bug-coreutils/2009-08/msg00048.html
This list archive has done strange things with character encodings which
make the discussion difficult to follow. Something along the
Philip Rowlands wrote:
On Mon, 7 Sep 2009, Eric Blake wrote:
This was discussed last month. The verdict is no.
http://lists.gnu.org/archive/html/bug-coreutils/2009-08/msg00048.html
This list archive has done strange things with character encodings
which make the discussion difficult to
Philip Rowlands wrote:
On Mon, 7 Sep 2009, Eric Blake wrote:
This was discussed last month. The verdict is no.
http://lists.gnu.org/archive/html/bug-coreutils/2009-08/msg00048.html
This list archive has done strange things with character encodings which
make the discussion difficult to
Pádraig Brady wrote 1276 bytes:
Also some scripts may be depending on the output from `cp -v`,
so I'm not on for changing it.
The output is already changed (the neat quotes). And one can use
LC_ALL=POSIX to get rid of both.
Pádraig Brady wrote:
Philip Rowlands wrote:
On Mon, 7 Sep 2009, Eric Blake wrote:
This was discussed last month. The verdict is no.
http://lists.gnu.org/archive/html/bug-coreutils/2009-08/msg00048.html
This list archive has done strange things with character encodings which
make the
Jim Meyering j...@meyering.net writes:
I considered that and discussed the
trade-off in the comment I committed.
There have been systems and configurations with
nonexistent and unusable /dev/stdin files.
sorry, I didn't read you comment.
This patch changes `tail' to handle stdin separately
Hi Jim,
On Mon, Sep 07, 2009 at 11:09:21AM +0200, Jim Meyering wrote:
There have been disproportionately many bug fixes since coreutils-7.5.
It's an interesting mix of fixes for recent regressions and for a few older
bugs.
coreutils snapshot:
http://meyering.net/cu/coreutils-ss.tar.gz
Erik Auerswald wrote:
...
http://meyering.net/cu/coreutils-7.5.65-61cc6.tar.gz
http://meyering.net/cu/coreutils-7.5.65-61cc6.tar.xz
Hi Erik,
I've run the non-root checks without failures on my debian/sid x86 (32 bit)
system.
Good to know. That's one I hadn't tested.
Thanks!
Giuseppe Scrivano wrote:
Jim Meyering j...@meyering.net writes:
I considered that and discussed the
trade-off in the comment I committed.
There have been systems and configurations with
nonexistent and unusable /dev/stdin files.
sorry, I didn't read you comment.
This patch changes `tail'
Pádraig Brady p...@draigbrady.com writes:
# quick dir listing with latest files/dirs at the bottom,
# prettify symlink arrows.
# using eval to precompute the tput sequences.
eval
l() {
ls -lrt --color=always \\...@\ |
sed 's/ - / $(tput bold)▪▶$(tput sgr0) /'
FWIW, I find this arrow
Hi all,
I'm failing in creating a symlink where the destination is a directory
and do a backup of this directory. This is what I tried:
$ mkdir a
$ ln -s --no-target-directory --backup=numbered b a
ln: `a': cannot overwrite directory
Looking at the code I can seen the reason, the backup
Andreas Schwab wrote:
Pádraig Brady p...@draigbrady.com writes:
# quick dir listing with latest files/dirs at the bottom,
# prettify symlink arrows.
# using eval to precompute the tput sequences.
eval
l() {
ls -lrt --color=always \\...@\ |
sed 's/ - / $(tput bold)▪▶$(tput sgr0) /'
Jim Meyering wrote:
There have been disproportionately many bug fixes since coreutils-7.5.
It's an interesting mix of fixes for recent regressions and for a few older
bugs.
Passed Skipped Failed
\-
Fedora core 5
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Eric Blake on 9/7/2009 10:30 AM:
Any objections to deleting the rename-dest-slash module? It performs a
subset of the rename module,
Correction. As currently written, rename.m4 checks whether:
rm -rf d1 d2
mkdir d1
rename(d1/,d2)
I've seen this test fail a few times, recently,
but it's not easy to trigger. With this change,
I saw no failure in 60 iterations.
From 494fed027114d63719439b399a7602f8d0384bcf Mon Sep 17 00:00:00 2001
From: Jim Meyering meyer...@redhat.com
Date: Mon, 7 Sep 2009 11:39:19 +0200
Subject: [PATCH]
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Pádraig Brady on 9/7/2009 10:22 AM:
Solaris 9 x86 build failed with:
fstatat.c, line 39: undefined symbol: AT_SYMLINK_NOFOLLOW
Hmm. That should be taken care of by the gnulib fcntl.h replacement.
Are we missing a #include? Can you
Pádraig Brady wrote:
Jim Meyering wrote:
There have been disproportionately many bug fixes since coreutils-7.5.
It's an interesting mix of fixes for recent regressions and for a few older
bugs.
Passed Skipped Failed
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Jim Meyering on 8/22/2009 1:58 AM:
John Taylor wrote:
I'm trying to build coreutils-7.4 with a cross-compilator (gcc-4.4.1)
that runs on i686-unknown-linux-gnu and produces code for
x86_64-unknown-linux-gnu. When building coreutils, I
Eric Blake wrote:
According to Pádraig Brady on 9/7/2009 10:22 AM:
Solaris 9 x86 build failed with:
fstatat.c, line 39: undefined symbol: AT_SYMLINK_NOFOLLOW
Hmm. That should be taken care of by the gnulib fcntl.h replacement.
Are we missing a #include? Can you post the full failure?
Eric Blake wrote:
Solaris 10 failures:
tail-2/flush-initial (patch attached)
Hmm. We should reword the NEWS entry (right now, it states that the
failure was rare because stdbuf was not always used; but your analysis of
this failure proves that it can indeed be more common).
Do you know of
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Jim Meyering on 9/7/2009 11:41 AM:
Eric Blake wrote:
Solaris 10 failures:
tail-2/flush-initial (patch attached)
Hmm. We should reword the NEWS entry (right now, it states that the
failure was rare because stdbuf was not always
Hi Pádraig,
On my 6th run of make check, the cat-buf test failed.
It's an inherently racy test.
If the dd process is starved until 2 is printed,
it will end up printing both lines to out.
Here's the code:
# Use a fifo rather than a pipe in the tests below
# so that the producer (cat)
While I was looking at this test...
From a4a864da365fe70eb3a69fd4347f8f747a258efd Mon Sep 17 00:00:00 2001
From: Jim Meyering meyer...@redhat.com
Date: Mon, 7 Sep 2009 20:23:03 +0200
Subject: [PATCH] tests: misc/cat-buf: clean up syntax
* tests/misc/cat-buf: Don't suppress dd's stderr.
Remove
My recent change in this area wasn't quite right.
For example, before,
: k; ./tail -f - k .
would revert to sleep-based implementation.
With this patch, since - ends up being ignored, the inotify-based
implementation is used.
From a8d26b3ce1630b6e9213b79d213ad7c699ee9861 Mon Sep 17
Ulrich Drepper wrote:
Jim Meyering wrote:
Thanks for the report!
Ulrich and Ren reported that the command,
echo foobar | tail -c3 -f
would block indefinitely, while POSIX says that it must not.
Here's how I expect to fix this.
Maybe I'm splitting hairs, but rather than
call this a bug in
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Jim Meyering on 9/7/2009 2:48 PM:
This seems small and safe enough that
I'm leaning toward including it in coreutils-7.6,
but I'll wait for a second opinion and/or review.
+** POSIX conformance
+
+ tail -f now ignores - when
Eric Blake wrote:
According to Jim Meyering on 9/7/2009 2:48 PM:
This seems small and safe enough that
I'm leaning toward including it in coreutils-7.6,
but I'll wait for a second opinion and/or review.
+** POSIX conformance
+
+ tail -f now ignores - when stdin is a pipe or FIFO, per
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Eric Blake on 9/7/2009 11:12 AM:
According to Eric Blake on 9/7/2009 10:30 AM:
Any objections to deleting the rename-dest-slash module? It performs a
subset of the rename module,
Correction. As currently written, rename.m4 checks
Jim Meyering wrote:
Hi Pádraig,
On my 6th run of make check, the cat-buf test failed.
It's an inherently racy test.
If the dd process is starved until 2 is printed,
it will end up printing both lines to out.
Hmm. 0.2s was too short between writes so.
I've bumped it up to 0.5s since 0.2
Jim Meyering wrote:
Ulrich Drepper wrote:
Jim Meyering wrote:
Thanks for the report!
Ulrich and Ren reported that the command,
echo foobar | tail -c3 -f
would block indefinitely, while POSIX says that it must not.
This seems to be handled already (see line 1926):
echo foobar |
Eric Blake wrote:
According to Pádraig Brady on 9/7/2009 10:22 AM:
Solaris 9 x86 build failed with:
fstatat.c, line 39: undefined symbol: AT_SYMLINK_NOFOLLOW
Hmm. That should be taken care of by the gnulib fcntl.h replacement.
Are we missing a #include? Can you post the full failure?
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Pádraig Brady on 9/7/2009 6:32 PM:
fstatat.c, line 39: undefined symbol: AT_SYMLINK_NOFOLLOW
Hmm. That should be taken care of by the gnulib fcntl.h replacement.
Are we missing a #include? Can you post the full failure?
gnulib
Pádraig Brady wrote:
Hmm. 0.2s was too short between writes so.
I've bumped it up to 0.5s since 0.2 failed so infrequently.
I've also changed to just skip on possible failure.
That should do it. Thanks.
53 matches
Mail list logo