I'd rather keep the open module alone. However, there's an interesting
thing that goes to our advantage. You cannot detect O_CLOEXEC at
build-time, otherwise binary compatibility with older kernels goes
south: I did this for GNU Smalltalk and two people reported that they
needed to reboot to
Yes, but unlike dup2.c where we are emulating fcntl(n, F_DUPFD, 3), you
are now introducing an arbitrarily large target
Actually, the same happens in dup2.c; the code I referred to is the one
emulating dup2 using dup. You are confusing with your own
popen-safer.c, which indeed has a very
This patch tests for the presence of O_CLOEXEC at run-time.
The technique I use allows to use only two system calls instead of
three on system where O_CLOEXEC does not work, at least after the
first call to open_noinherit.
The code can be moved to an emulation of open by doing something
like:
If I touch lib/glob.in.h and create an unwritable named lib/glob.h-t,
that causes all subsequent build attempts to fail, because redirecting
to an unreadable file fails. (that actually happens if you happen to
experiment with an overly restrictive umask)
There are several modules that have this
Bruno Haible br...@clisp.org writes:
I don't think there's much point in spending time on these failures.
Ultimately people want to use Win32 binaries on Windows (or possibly
ReactOS), but not on Wine. People wanting to execute programs on Linux
will use the ELF format compiled programs, not
Jim Meyering wrote:
that actually happens if you happen to
experiment with an overly restrictive umask
No one of us did such experiments. Which is why it didn't come up
earlier.
+ consistently remove $...@-t before redirecting to it
+ * modules/argz: Remove $...@-t and $@ before
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Paolo Bonzini on 8/21/2009 1:34 AM:
(or, is mingw subject to a compile-time maximum of open fds
where we can just return EMFILE up front?).
EMFILE is more appropriate that EBADF; POSIX states this for fcntl:
[EMFILE]
The argument
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Paolo Bonzini on 8/21/2009 1:42 AM:
This patch tests for the presence of O_CLOEXEC at run-time.
The technique I use allows to use only two system calls instead of
three on system where O_CLOEXEC does not work, at least after the
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Jim Meyering on 8/21/2009 3:28 AM:
And with the following patch, if you add this line
to your project's configure.ac file,
AM_SILENT_RULES([yes]) # make --enable-silent-rules the default.
Or you can populate your config.site
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Simon Josefsson on 8/20/2009 7:22 AM:
Pushed. See
http://autobuild.josefsson.org/gnulib/log-200908200817225767000.txt
+
+ * tests/test-exclude1.sh: Handle Windows EOL.
Rather than adding a tr process, would it instead make sense
Hi!
Bison's bootstrap has a few changes compared to gnulib's. I'd like to
use the stock bootstrap from gnulib, so there are a few changes that
we would like to submit.
First, a trivial matter: 'bootstrap | less' does not work as expected.
From b490a83f8ba3f63f1a43728140c80ec924f1f552 Mon
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Bruno Haible on 8/20/2009 2:14 AM:
This is becoming confusing: Why should the 'fopen' test suddenly test
fopen_safer,
not fopen??
I would find it better to introduce a new test module fopen-safer-tests, that
shares
the bulk of
Akim Demaille wrote:
Hi!
Bison's bootstrap has a few changes compared to gnulib's. I'd like to
use the stock bootstrap from gnulib, so there are a few changes that
we would like to submit.
First, a trivial matter: 'bootstrap | less' does not work as expected.
From
On Fri, Aug 14, 2009 at 01:38:39PM +0200, Tom G. Christensen wrote:
Unfortunately another one just turned up:
gcc -std=gnu99 -g -O2 -L/usr/tgcware/lib -Wl,-R,/usr/tgcware/lib
-Wl,-z,ignore
-o test-getopt test-getopt.o ../gllib/libgnu.a -lm
Undefined first referenced
Ok for mine (probably only modules/poll) too.
I post this also here, cause it seems a gnulib error
This is about the check for Coreutils 7.5
As you know Os2 has a problem with the check,
so i did a check on the lib first
make -k -C gnulib-tests check gnulib-tests.log 21
But i see errors about mmap
CC test-memchr.o
CCLD
Eric Blake wrote:
According to Jim Meyering on 8/21/2009 3:28 AM:
And with the following patch, if you add this line
to your project's configure.ac file,
AM_SILENT_RULES([yes]) # make --enable-silent-rules the default.
Or you can populate your config.site file with:
Eric Blake wrote:
+ * tests/test-exclude1.sh: Handle Windows EOL.
Rather than adding a tr process, would it instead make sense to make
test-exclude.c use the binary-io module to ensure binary output, so that
no \r are generated in the first place?
The technique of piping through 'tr' is
18 matches
Mail list logo