Jim Meyering [Sun, 21 Oct 2007 10:38:33 +0000 (12:38 +0200)]
tr: do not reject an unmatched [:lower:] or [:upper:] in SET1.
* NEWS: Mention this.
* src/tr.c (get_next) [RE_CHAR_CLASS]: Don't skip the loop when
processing [:lower:] and [:upper:].
(main): Require [:lower:] or [:upper:] in SET1 only when
when one of those is specified in SET2.
* tests/tr/Test.pm: Add tests for this fix.
Reported by Per Starbäck.
Jim Meyering [Sat, 20 Oct 2007 12:48:29 +0000 (14:48 +0200)]
Detect printf(3) failure due to ENOMEM.
* src/printf.c: Include "xprintf.h"
(print_direc): Use xprintf, rather than printf.
* bootstrap.conf (gnulib_modules): Add xprintf.
* po/POTFILES.in: Add lib/xprintf.c.
Jim Meyering [Sat, 20 Oct 2007 08:07:12 +0000 (10:07 +0200)]
Adjust tests to pass, now that Test.pm ones use check.mk.
* tests/check.mk (vc_exe_in_TESTS): Adjust regexp not to match the
"TESTS = $x-tests" line in each Test.pm-derived Makefile.am file.
* tests/sort/main: Remove this unused file.
Jim Meyering [Tue, 9 Oct 2007 15:12:45 +0000 (17:12 +0200)]
Move the help-version test into misc/.
* tests/help-version: Move to...
* tests/misc/help-version: ...here.
Source test-lib.sh, as usual.
Use $abs_top_builddir, rather than ".." and "../..".
Jim Meyering [Tue, 9 Oct 2007 07:58:27 +0000 (09:58 +0200)]
Skip the chcon test on a system with no SELinux support.
* tests/test-lib.sh (require_selinux_): New function.
* tests/misc/chcon: Use it.
* tests/misc/selinux: Use it here, too.
* tests/cp/cp-a-selinux: and here.
* tests/selinux: Remove file.
* tests/Makefile.am (EXTRA_DIST): Remove selinux.
Reported by Mike Frysinger and Bauke Jan Douma.
Jim Meyering [Mon, 8 Oct 2007 08:34:04 +0000 (10:34 +0200)]
When forming a truncated name, use the "[...]" prefix even for
an extremely long name in the current directory.
* src/remove.c (full_filename_): Rename a local variable.
Use two separate "truncated" variables, in case the second
call to right_justify does not indicate any truncation -- which
would happen only if the single component FILENAME were longer
than 511 bytes and DIR_NAME were empty.
Jim Meyering [Sun, 7 Oct 2007 20:58:29 +0000 (22:58 +0200)]
rm could malfunction under unusual circumstances:
When operating on a relative name longer than 511 bytes,
and (when either processing a directory that is neither writable
nor readable (but still searchable) or when determining whether
to prompt), and encountering an ENOMEM error while forming the
file name, rm would operate on a truncated-to-511-byte name
starting with "[...]" rather than the intended one.
* NEWS: Describe the bugs.
* src/remove.c: Correct two misuses of full_filename:
(full_filename0, xfull_filename): New functions.
(full_filename_): Rewrite to use full_filename0.
(AD_pop_and_chdir): Use xfull_filename, not full_filename.
(write_protected_non_symlink): Likewise.
Jim Meyering [Sat, 2 Jun 2007 21:04:10 +0000 (23:04 +0200)]
New program: mktemp.
* NEWS: Mention this.
* README: Add mktemp to the list.
* AUTHORS: Add this: mktemp: Jim Meyering
* src/mktemp.c: New file.
* src/Makefile.am (bin_PROGRAMS): Add mktemp.
(mktemp_LDADD): Add $(LIB_GETHRXTIME).
* man/mktemp.x: New file.
* man/Makefile.am (dist_man_MANS): Add mktemp.1.
(mktemp.1): New dependency.
* man/.cvsignore: Add mktemp.1.
* man/.gitignore: New file.
* src/.cvsignore, src/.gitignore: Add mktemp.
* tests/misc/mktemp: New file.
* tests/misc/Makefile.am (TESTS): Add mktemp.
* tests/Coreutils.pm (run_tests): Give the POST-test function
access to stdout and stderr contents, so it can verify that
the named-on-stdout file/dir does indeed exist and has proper
permissions, etc.
[po/ChangeLog]
* POTFILES.in: Add src/mktemp.c.
Jim Meyering [Sat, 1 Sep 2007 08:33:33 +0000 (10:33 +0200)]
Make tempname more random, via the randint module.
* gl/modules/tempname (Depends-on): Add randint and stdbool.
* gl/lib/tempname.c: Include randint.h and stdbool.h.
(uint64_t): Remove definition. Not needed.
[_LIBC] (RANDOM_BITS): Remove this block, now that we have proper random bits.
(check_x_suffix): New function.
(gen_tempname_len): Rename from __gen_tempname.
Add a parameter, x_suffix_len, telling how many X's there must be at
the end of the template.
Use pseudo-random numbers all the way, rather than adding 7777
from one iteration to the next.
(__gen_tempname): New function, to call gen_tempname_len, requiring a
suffix length of 6.
* gl/lib/tempname.h: Add prototype for gen_tempname_len.
Jim Meyering [Sat, 1 Sep 2007 07:54:45 +0000 (09:54 +0200)]
Convert coreutils' rand*.{c,h,m4} into modules.
First step: move these files to gl/lib:
* lib/rand-isaac.c, lib/rand-isaac.h
* lib/randint.c, lib/randint.h
* lib/randperm.c, lib/randperm.h
* lib/randread.c, lib/randread.h
Step 2: add modules/rand* and remove now-unneeded .m4 files.
* gl/modules/randint: New file.
* gl/modules/randperm: New file.
* gl/modules/randread: New file.
* m4/randint.m4: Remove file.
* m4/randperm.m4: Remove file.
* m4/randread.m4: Remove file.
Step 3: use the new modules
* bootstrap.conf (gnulib_modules): Add randint and randperm.
* m4/prereq.m4 (gl_RANDINT, gl_RANDREAD, gl_RANDPERM): Don't require;
These have been removed.
(gl_ROOT_DEV_INO): Don't require; already handled via bootstrap.conf.
Jim Meyering [Fri, 5 Oct 2007 08:55:26 +0000 (10:55 +0200)]
Make a failing cross-partition mv give a sensible diagnostic.
A cross-partition move of a file in a sticky tmpdir and owned by
another user would evoke an invalid diagnostic after copying it:
mv: cannot remove `x': Operation not permitted
Either of the following (mv.c, remove.c) changes would fix the bug by
itself. I think it's slightly better to use both; the added cost is
minimal: mv: an extra lstat-per-mv-cmdline-arg-that-goes-cross-partition,
rm: an extra lstat-per-unlink-that-fails-w/EPERM.
* src/remove.c (remove_entry): Also lstat the file upon EPERM.
* src/mv.c (rm_option_init): Initialize root_dev_ino just as is done
in rm, so that a cross-partition invoked remove.c:rm call works the
same way as one invoked from the command-line use of "rm". That
setting of root_dev_ino makes rm() do the equivalent of an additional
lstat for each argument, which in turn gives rm enough information to
issue the right diagnostic.
* tests/mv/sticky-to-xpart (version): New file. Test for the above.
* tests/mv/Makefile.am (TESTS): Add sticky-to-xpart.
Arrange for "make check-root" to run the new root-only test.
* tests/Makefile.am (tb): New target, to run the new root-only test.
(all_t): Add tb.
* src/c99-to-c89.diff: Adjust offsets.
Jim Meyering [Fri, 5 Oct 2007 10:14:16 +0000 (12:14 +0200)]
Add PACKAGE_VERSION to TESTS_ENVIRONMENT via check.mk.
* tests/check.mk (TESTS_ENVIRONMENT): Add PACKAGE_VERSION here,
rather than in every Makefile.am that needs it.
* tests/rm/Makefile.am (TESTS_ENVIRONMENT): Remove PACKAGE_VERSION.
* tests/misc/Makefile.am (TESTS_ENVIRONMENT): Likewise.
Jim Meyering [Tue, 2 Oct 2007 09:10:22 +0000 (11:10 +0200)]
Never copy through a symlink that cp has just created.
* src/copy.c (copy_internal): When same-file detection requires
'stat'ing the destination file, also 'lstat' it and ensure that
it wasn't the destination of a preceding copy operation.
This bug was introduced on 2007-06-18.
* tests/cp/abuse: New test for the above.
* tests/cp/Makefile.am (TESTS): Add abuse.
Jim Meyering [Tue, 25 Sep 2007 20:24:21 +0000 (22:24 +0200)]
Move file-set and hash-triple modules to gnulib.
* bootstrap.conf (gnulib_modules): Remove file-set, now that
it's in gnulib, and the canonicalize module requires it there.
* gl/lib/file-set.c, gl/lib/file-set.h, gl/modules/hash-triple: Remove.
* gl/lib/hash-triple.c, gl/lib/hash-triple.h, gl/modules/file-set: Remove.
Jim Meyering [Mon, 24 Sep 2007 07:53:57 +0000 (09:53 +0200)]
Add a test to exercise a readlink bug.
* tests/misc/readlink-fp-loop: New file. Test for the readlink bug
fixed through today's change to Gnulib's canonicalize module.
* tests/misc/Makefile.am (TESTS): Add readlink-fp-loop.
Bug report and a test case from mpb.mail@gmail.com.
Make "Date input formats" easier to find in info file
* doc/coreutils.texi (date invocation):
`man date` gives all options for the date command except
the description of the date input formats. Therefore it
should be easy to browse to this info using `info date`.
Jim Meyering [Tue, 25 Sep 2007 19:27:14 +0000 (21:27 +0200)]
Use XOR, not OR to combine bits.
* gl/lib/hash-triple.c (triple_hash): Use XOR (^), not OR (|), to
combine the bits from hashing the name and those of the inode number.
Add a few comments and remove out-of-context ones.
Jim Meyering [Sat, 22 Sep 2007 08:02:09 +0000 (10:02 +0200)]
rm: give a sensible diagnostic when failing to remove a symlink
On some systems (those with openat et al), when rm would fail to
remove a symlink, it would fail with the misleading diagnostic,
"Too many levels of symbolic links".
* NEWS: Mention the bug fix.
* src/remove.c (is_nondir_lstat): New function.
(remove_entry): Use it to catch failed-to-remove symlink (and any
other non-dir) here so that we don't fall through and try to treat
it as directory, which -- with a symlink -- would provoke the bogus
ELOOP failure.
* tests/rm/fail-eacces: Add a test for the above.
* src/c99-to-c89.diff: Adjust offsets.
Jim Meyering [Sat, 22 Sep 2007 06:40:47 +0000 (08:40 +0200)]
rm: fix a tiny, nearly inconsequential bug.
Don't perform a "."-relative lstat, when the file in question
may well not be in ".". Although this is a bug, a few attempts
to exercise it on a linux-2.6.22 system failed. You probably need
a pre-openat system to trigger the failure. The consequence of this
bug would be a lower-quality diagnostic upon failed dir removal.
* src/remove.c (is_dir_lstat): Add a parameter, fd_cwd.
Use it instead of hard-coding AT_FDCWD.
(remove_entry): Call is_dir_lstat with fd_cwd.
Jim Meyering [Thu, 13 Sep 2007 10:13:35 +0000 (12:13 +0200)]
Enable some previously omitted test scripts. Add a test cross-check.
* tests/rm/Makefile.am (TESTS): Add empty-name and unreadable.
* tests/rm/empty-name: Likewise.
* tests/rm/unreadable: Reenable this test. Adjust for new rm.
* tests/mkdir/Makefile.am: Add writable-under-readonly.
* tests/mkdir/writable-under-readonly: Add some comments.
This test is always skipped, for now.
* tests/Makefile.am (ta): Hook up the new root-only script.
* tests/tail-2/infloop-1: Make this test pass.
* tests/tail-2/Makefile.am (TESTS): Add infloop-1.
* tests/tail-2/fflush: Remove unused file.
* tests/check.mk (vc_executable_is_in_TESTS): More portable.
* tests/check.mk (check): Depend on the above.
* build-aux/check.mk: Remove comment mentioning AUTHORS file.
Jim Meyering [Thu, 13 Sep 2007 08:04:35 +0000 (10:04 +0200)]
Factor out definitions in TESTS_ENVIRONMENT of srcdir, top_srcdir, etc.
tests/check.mk (top_srcdir): Define.
tests/*/Makefile.am: Remove definitions of $(srcdir), $(top_srcdir),
$(abs_top_srcdir), and $(abs_top_builddir), since they're
defined via the included tests/check.mk.
* tests/misc/Makefile.am (TESTS): Add the new files.
* tests/Makefile.am (SUBDIRS): Remove the dir names.
* tests/od, tests/sha1sum, tests/shred, tests/stty: Remove the directories.
* configure.ac (AC_CONFIG_FILES): Remove the corresponding Makefile names.
Jim Meyering [Tue, 11 Sep 2007 21:30:09 +0000 (23:30 +0200)]
Move the two tests in tests/tee to tests/misc/.
* tests/tee/basic: Move this file to ...
* tests/misc/tee: ...here. Don't rely on $PROG in env.
* tests/tee/dash: Move this file to ...
* tests/misc/tee-dash: ...here. Don't rely on $PROG in env.
* tests/misc/Makefile.am (TESTS): Add tee.
* tests/Makefile.am (SUBDIRS): Remove tee.
* tests/tee: Remove the directory.
* configure.ac (AC_CONFIG_FILES): Remove tests/tee/Makefile
Jim Meyering [Mon, 10 Sep 2007 22:43:05 +0000 (00:43 +0200)]
Move the two tests in tests/sum/ to tests/misc/.
* tests/sum/basic-1: Move this file to ...
* tests/misc/sum: ...here. Don't rely on $PROG in env.
* tests/sum/sysv: Move this file to ...
* tests/misc/sum-sysv: ...here. Adapt it to use test-lib.sh.
* tests/misc/Makefile.am (TESTS): Add sum and sum-sysv.
* tests/Makefile.am (SUBDIRS): Remove sum.
* tests/sum: Remove the directory.
* configure.ac (AC_CONFIG_FILES): Remove tests/sum/Makefile.
Jim Meyering [Thu, 6 Sep 2007 21:05:16 +0000 (23:05 +0200)]
Convert tests/mv, too.
* tests/other-fs-tmpdir: Before, all callers would exit 77 upon failure to
find the required dir. Now, exit 77 in this script so callers don't have to.
Adjust callers.
Jim Meyering [Thu, 16 Aug 2007 14:28:11 +0000 (16:28 +0200)]
Parallel "make check" support.
* build-aux/check.mk: New file, from The Vaucanson Group.
* .x-sc_GPL_version: New file, to allow "version 2 or later"
in build-aux/check.mk.
* Makefile.am (EXTRA_DIST): Add .x-sc_GPL_version.
* tests/check.mk: New file.
* tests/Makefile.am (EXTRA_DIST): Add check.mk, mkdtemp and test-lib.sh.
Begin factoring "sample-test" out of test scripts.
* tests/test-lib.sh: New file, to be sourced by all tests that
were previously derived from the "sample-test" template.
* tests/mkdtemp: New file.
* tests/touch/dir-1: Use test-lib.sh.
* tests/touch/empty-file: Likewise.
* tests/touch/fail-diag: Likewise.
* tests/touch/fifo: Likewise.
* tests/touch/no-create-missing: Likewise.
* tests/touch/no-rights: Likewise. Also, don't sleep.
* tests/touch/not-owner: Likewise.
* tests/touch/obsolescent: Likewise.
* tests/touch/read-only: Likewise.
* tests/touch/relative: Likewise.
* tests/touch/Makefile.am: Include $(top_srcdir)/tests/check.mk,
to get the parallel-"make check" bits.
Move a slow test into tests/misc.
* tests/check.mk: Wrapper.
* tests/ls/time-1: Move this file to tests/misc/ls-time.
* tests/misc/ls-time: New file. From tests/ls/time-1.
* tests/ls/Makefile.am (TESTS): Remove time-1.
* tests/misc/Makefile.am (TESTS): Add ls-time.
Jim Meyering [Fri, 7 Sep 2007 08:37:08 +0000 (10:37 +0200)]
chmod: don't ignore a dangling symlink
* NEWS: Mention the bug fix.
* src/chmod.c (process_file): Handle the case of FTS_SLNONE,
i.e., give a diagnostic saying we cannot operate on such a file.
* tests/chmod/thru-dangling: Compare new stderr output with expected.
Jim Meyering [Wed, 5 Sep 2007 08:57:54 +0000 (10:57 +0200)]
Adapt to new SELinux behavior: "?" vs. new "unlabeled"
* src/ls.c (gobble_file): Interpret the new "unlabeled" indicator
from getfilecon/lgetfilecon the same way we interpret a negative
return value: no security context. So we don't print the "+".
* tests/selinux: Recognize that "unlabeled" means insufficient
support for SELinux, just like "?".