Jim Meyering [Sun, 2 Dec 2007 20:02:26 +0000 (21:02 +0100)]
Copy a .po file into place also when the target doesn't exist.
* bootstrap (gnulib_tool): ... not just when the checksum differs.
So now, if I manually remove po/xx.po, rerunning bootstrap will
copy the latest into place.
Jim Meyering [Sun, 2 Dec 2007 19:53:43 +0000 (20:53 +0100)]
Make my-distcheck depend on "check".
* Makefile.maint (my-distcheck): Also depend on "check", so I can't tag
and release something that fails "make check" (however harmless) when
run in a checked-out-from-git dir, but not in the distributed tarball.
Jim Meyering [Sun, 2 Dec 2007 15:19:04 +0000 (16:19 +0100)]
Remove just-installed su, if unable to make it set-UID root.
Normally, su isn't even installed. However, if you configure with
--enable-install-program=su, and then install with insufficient
privileges, we now make sure to remove the just-installed binary.
* src/Makefile.am (install-exec-hook): Rename from install-exec-local,
so that this rule is guaranteed to be run *after* installation.
If unable to chown and chmod the installed "su" program, remove it.
Reported by Greg Schaefer.
Jim Meyering [Sat, 1 Dec 2007 18:52:45 +0000 (19:52 +0100)]
Change release procedure: tag *before* running e.g., "make stable".
* Makefile.maint (vc-dist): Don't tag. Now, you must apply the tag
before running "make beta", "make stable", etc.
(vc-tag-check): Remove rule. No longer makes sense.
Jim Meyering [Sat, 1 Dec 2007 17:50:40 +0000 (18:50 +0100)]
Defer NEWS and ChangeLog checks until after the version-changing tag.
* Makefile.maint (alpha beta major): Don't depend on the
news-date-check changelog-check targets. They must fail before
applying the version-changing tag. Instead, run them after "vc-dist".
Paul Eggert [Sat, 1 Dec 2007 09:10:52 +0000 (10:10 +0100)]
Add a test for cp -p and existing file permissions.
* tests/cp/Makefile.am (TESTS): Add existing-perm-race.
* tests/cp/existing-perm-race: New test. It isn't much of a
test yet, since it's hard to catch the race, but it has a FIXME
that will let us do a better test later.
Paul Eggert [Sat, 1 Dec 2007 09:09:57 +0000 (10:09 +0100)]
Fix a security race with "cp -p A B" when B already exists.
* src/copy.h (struct cp_options): New member owner_privileges.
* src/copy.c (USE_ACL): Define to 0 if not defined, for convenience.
(owner_failure_ok): New function.
(set_owner): Avoid a security-related race by doing an extra chmod
first if it looks like there might be trouble right after a chown.
Accept a source struct stat rather than a uid and gid, and
accept a boolean NEW_DST and destination struct stat.
All callers changed.
* src/copy.h (cp_options_default): New function, replacing the
old chown_privileges.
* src/copy.c (cp_options_default): Likewise.
* src/cp.c (cp_option_init): Use it.
* src/install.c (cp_option_init): Likewise.
* src/mv.c (cp_option_init): Likewise.
Jim Meyering [Fri, 30 Nov 2007 12:45:38 +0000 (13:45 +0100)]
Be extra careful to quote $abs_top_builddir-derived names.
* tests/misc/ls-misc (shell_quote): New function.
Use it to quote file names derived from $abs_top_builddir,
in case it contains shell meta-characters. This is not currently
needed, since CuTmpdir detects the fishy name and skips the test.
But it's important enough to add the extra protection.
Reported by Ralf Wildenhues.
Ralf Wildenhues [Fri, 30 Nov 2007 07:53:38 +0000 (08:53 +0100)]
* tests/CuTmpdir.pm (END): Do not do the cleanup if $dir is not
defined, because it then goes wild and changes the mode of all
directories below $HOME. Undefined $dir can happen if the test
is to be skipped because of an unsafe working directory name.
Jim Meyering [Tue, 27 Nov 2007 18:26:25 +0000 (19:26 +0100)]
Fix md5sum+sha*sum on sparc: regenerate files with latest autoconf.
* configure et al: Regenerate with very latest version of autoconf
(v2.61a-312-gb524b0f), to fix md5sum and sha*sum miscompilation
on big-endian systems (like sparc) due to lack of the required
definition in of WORDS_BIGENDIAN in lib/config.h.
Jim Meyering [Mon, 26 Nov 2007 15:34:58 +0000 (16:34 +0100)]
setuidgid.c: more cleanup
* src/setuidgid.c: Include "mgetgroups.h".
(main): Don't presume there's a fixed limit on the maximum number
of group IDs. Don't use NGROUPS; it's artificially low.
Change the name of a local: s/gids_count/n_gids/.
Remove non-portable "const": s/const int tmp =.../int tmp =.../
Lasse Collin [Fri, 9 Nov 2007 23:07:37 +0000 (01:07 +0200)]
"cp -p" tries to preserve GID even if preserving the UID fails.
* NEWS: Mention this new feature.
* src/copy.c (set_owner): Try to preserve just the GID,
when initial fchown/lchown fails.
* src/cp.c (re_protect): Likewise.
Jim Meyering [Thu, 22 Nov 2007 07:42:32 +0000 (08:42 +0100)]
Make it easier to diagnose PATH-induced "make distcheck" failures.
* Makefile.maint (write_loser): Define.
(my-distcheck): Use an always-failing wrapper script, that gives
a diagnostic, not "false".
Run only "make -C tests check" with the restrictive PATH.
Run the new gnulib-tests/ with the usual PATH.
Jim Meyering [Fri, 16 Nov 2007 08:31:15 +0000 (09:31 +0100)]
cp: by default, refuse to copy through a dangling destination symlink
* NEWS: Mention this change.
* doc/coreutils.texi (cp invocation): Describe the new behavior.
* src/copy.c: No longer include "canonicalize.h".
(copy_reg): Upon failure to open a dangling destination symlink, don't
canonicalize the name, but rather fail (default) or, with POSIXLY_CORRECT,
repeat the open call without O_EXCL (potentially dangerous).
* src/copy.h (struct cp_options) [open_dangling_dest_symlink]:
New member. Reorder the others, grouping "bool" and "enum"
members together.
* tests/cp/thru-dangling: Test for changed and new behavior.
* src/cp.c (cp_option_init): Initialize new member.
* src/install.c (cp_option_init): Likewise.
* src/mv.c (cp_option_init): Likewise.
Paul Eggert [Fri, 16 Nov 2007 23:04:08 +0000 (00:04 +0100)]
Port tests/rmdir/ignore away from GNU/Linux.
* tests/rmdir/ignore: Don't assume that rmdir($PWD) will fail
with errno==ENOTEMPTY when $PWD is not empty; Posix also allows
rmdir to fail with errno==EBUSY.
Paul Eggert [Fri, 16 Nov 2007 10:04:31 +0000 (11:04 +0100)]
Port readlink-fp-loop to Solaris.
* tests/misc/readlink-fp-loop (symlink_loop_msg): New var,
which records the symlink-loop message, whose wording is
not standardized by Posix. Do not rely on "echo x > p/1"
to work when p/1 has a lot of indirect symlinks. (I'm surprised
that it works on Linux. Perhaps a Linux bug?)
Jim Meyering [Fri, 16 Nov 2007 10:03:03 +0000 (11:03 +0100)]
Port to Solaris 'make' and use a Posixish shell on Solaris.
* bootstrap.conf (gnulib_modules): Add gnu-make, posix-shell.
* build-aux/check.mk (SHELL): Set to $(PREFERABLY_POSIX_SHELL),
so that commands can assume Posix syntax.
(ENABLE_HARD_ERRORS, TEST_LOGS): Don't use GNU Make's "?=" syntax.
(SH_E_WORKAROUND): New macro.
(am__check_pre, $(TEST_SUITE_LOG)): Use it.
(am__check_pre): Fail if "mkdir" fails. Use $(SHELL)
rather than relying on the "#!/bin/sh" in the file, so that tests
can use Posix syntax.
(am__check_pre, am__tty_colors): Use $$src rather than $$<, to
support the Posix-make $(TEST_LOGS) rule.
(%.log: %.test, %.log: %$(EXEEXT)): Remove unused inference rules
that rely on a GNU Make extension and cause Solaris 'make' to fail.
(SUFFIXES): New macro, so that we can use Posix style inference rules.
(%.log: %): Use this rule only if GNU_MAKE.
Set $$src so that macros can use $$src rather than $$<.
(CHECK-FORCE, DEPENDENCY, $(TEST_LOGS)): New macros and rules,
which rely only on Posix 'make' semantics, and are used only with
non-GNU 'make' implementations. $(TEST_LOGS) invokes 'make'
recursively (and a bit inefficiently) to simulate the GNU 'make'
rules.
(.log.html): Renamed from "%.html: %.log", so that it relies only
on Posix 'make' semantics.
(check-clean, .PHONY): Do not depend on check-clean-local, since
Solaris 'make' complains about nonexistent rules like that.
* src/Makefile.am (SUFFIXES): Remove; no longer needed.
(groups): Use a specific rule rather than an inference rule that
is only instantiated once. The inference-rule approach does not
work with Solaris 'make', which gets confused by the "groups:
Makefile" line. It's not clear from the Posix spec that Solaris
'make' is buggy here, so instead of worrying about it, rewrite
the makefile so that it clearly conforms to Posix.
* tests/check.mk (TESTS_ENVIRONMENT): Export PACKAGE_BUGREPORT.
GNU 'make' does this automatically for us, but Solaris 'make'
doesn't.
Jim Meyering [Thu, 15 Nov 2007 12:04:13 +0000 (13:04 +0100)]
Avoid misbehavior of a cross-device "mv" or "install".
Those programs must not dereference a destination symlink.
* src/copy.c (copy_reg): Don't treat a dangling destination symlink
differently in move mode. In move mode, the only way the added
O_EXCL can cause failure is when some other process has recreated
the file this code unlinked a few instructions before.
Jim Meyering [Sat, 10 Nov 2007 17:33:07 +0000 (18:33 +0100)]
Enable gnulib-tool's --with-tests option.
* bootstrap: After all is done, transform the generated gnulib-tests/gnulib.mk.
* bootstrap.conf (gnulib_tool_option_extras): Set it here.
* gnulib-tests/Makefile.am: New file.
* Makefile.am (SUBDIRS): Add gnulib-tests.
* configure.ac (AC_CONFIG_FILES): Add gnulib-tests/Makefile.
* m4/jm-macros.m4 (coreutils_MACROS) [ARGMATCH_DIE_DECL]: Omit the
"extern " prefix to "void usage ()", so that the macro string value can
be used to construct a function definition in gnulib's test-argmatch.c.
* src/cp.c (usage) [-a]: Remove unnecessary 'P'.
Since option -d of 'cp' includes the function of -P, mentioning the
latter in the description of option -a is redundant and therefore
slightly confusing, as the reader tries to figure out what it adds.
Jim Meyering [Sat, 10 Nov 2007 16:31:31 +0000 (17:31 +0100)]
install+SELinux: reduce a 12x performance hit to ~1.5x
* src/install.c (setdefaultfilecon): Call matchpathcon_init_prefix,
to mitigate what would otherwise be a large performance hit due to
the use of matchpathcon.
Dan Walsh suggested the use of matchpathcon_init_prefix.
* gl/lib/se-selinux.in.h (matchpathcon_init_prefix): Define.
Jim Meyering [Mon, 5 Nov 2007 10:03:55 +0000 (11:03 +0100)]
Don't use GNU-sed-specific \< \> operators.
* configure.ac (MAN): Transform ginstall.1 to install.1 using
a loop, rather than sed with \<...\>.
Problem reported by Bruno Haible.
Andreas Schwab reminded me that \< and \> are not portable.
Paul Eggert [Sat, 3 Nov 2007 08:10:59 +0000 (01:10 -0700)]
Fix bug with "seq 10.8 0.1 10.95", plus another bug with %% in format.
* NEWS: Mention the %%-in-format bug fix.
* src/seq.c (struct layout): New type.
(long_double_format): New arg LAYOUT. Fill it in. Fix mishandling
of %% in formats.
(print_numbers): New arg LAYOUT. Don't convert LAST to output format
when deciding whether to go slightly past LAST. Instead, convert
X to output format and back. This fixes a bug reported by
Andreas Schwab in
<http://lists.gnu.org/archive/html/bug-coreutils/2007-10/msg00237.html>
where "seq 10.8 0.1 10.95" would output 11.0 on platforms where
10.95 rounds to a value that prints as 11.0 when only one digit
past the decimal point is asked for.
(main): Compute layout, for benefit of print_numbers.
* tests/misc/seq (float-3): Undo previous change, since the bug
should be fixed now.
(fmt-b): New test, for the %% bug.
Jim Meyering [Thu, 1 Nov 2007 17:58:17 +0000 (18:58 +0100)]
Use mktemp, not mkdtemp, to create test directories.
* tests/test-lib.sh: Use the mktemp binary we've just built,
not the mkdtemp script.
* tests/mkdtemp: Remove file.
* tests/Makefile.am (EXTRA_DIST): Remove mkdtemp.
Jim Meyering [Thu, 1 Nov 2007 11:10:06 +0000 (12:10 +0100)]
Adjust a seq subtest not to depend on the vagaries of floating point.
* tests/misc/seq (float-3): Use 10.94 as the endpoint, not 10.95,
since 10.95 was precisely in the middle of the interval, and with
a %.1f format could map to either 10.9 or 11.0.
Reported by Mike Frysinger
Jim Meyering [Thu, 1 Nov 2007 11:06:11 +0000 (12:06 +0100)]
Make the new printf-surprise test more precise.
* tests/test-lib.sh (require_ulimit_): New function.
* tests/misc/printf-surprise: Use ulimit -v to trigger the fixed bug,
and rather than checking printf's exit status (which would go wrong
on FreeBSD 6.1, since their printf(3) function doesn't require
lots of memory in this case) simply test whether it outputs
the first 10 bytes.
* tests/cp/same-file: Detect when linking to a symlink links to
the target of the symlink (FreeBSD 6.1 does this, Linux does not),
and skip the few tests that would otherwise fail.
Redirect output of final comparison to stderr, since all stdout
is already redirected.
Jim Meyering [Mon, 29 Oct 2007 08:47:52 +0000 (09:47 +0100)]
If $TERM is empty, don't use colors. This helps the buildbot,
since it produces its result in an environment with a tty, but
we don't want color codes in that case.
* build-aux/check.mk (am__tty_colors): Test for nonempty $TERM.
projects / thirdparty / coreutils.git / log
