Bernhard Voelker [Sat, 27 Oct 2012 11:24:18 +0000 (13:24 +0200)]
tests: shorten factor's inordinately-long test file names
Besides what the subject says, this commit moves the test data
for the factor tests from tests/local.mk into the directory
tests/factor/ where it belongs.
* tests/local.mk (EXTRA_DIST): Add new tests/factor/create-test.sh.
(p,q,t1,t2) Factor out the factor-related magic numbers.
(factor_tests): Rename the test names to t{00..36}.sh. Factor out
the triples of test data.
($(factor_tests)): Add dependency to new
tests/factor/create-factor.sh.
Call that script to generate the test scripts.
* tests/factor/run.sh: Turn this script into a template, and
therefore remove it's executable permission bit.
Add template variables START, END and CKSUM, replacing the code
to split the test data from the test script's file name.
Use the new template variables in the call to seq and for
creating the exp file.
* tests/factor/create-test.sh: Add new script to create the
test scripts from the template tests/factor/run.sh.
Use test data and magic numbers factored out from the above
files. Let the script also change the __TEMPLATE__ line in
run.sh to make clear that the test scripts are generated.
* cfg.mk (sc_tests_list_consistency): Exempt the new test.
(exclude_file_name_regexp--sc_prohibit_test_backticks):
Likewise.
Improved-by: Stefano Lattarini
Improved by: Jim Meyering
build: man page generation: degrade gracefully when perl is lacking
Since commit v8.19-118-g57da212, our 'dist-hook' rule tweaked the
distributed Makefile.in to make each man page 'man/foo.1' depend
on the corresponding source 'src/foo.c' rather than the corresponding
program 'src/foo'. That was done to accommodate systems without
perl, which cannot run help2man to regenerate the manpage after
its corresponding program has been built.
This seems a right and proper graceful degradation, in that the
man pages dependencies are still 100% correct in a git checkout,
while being more lax but "more portable" in a distribution tarball.
Alas, that is not the case in practice, as it turns out the tweaked
Makefile makes the building of man pages unreliable and potentially
incorrect!
In fact, assume that instead of the correct a dependency:
man/ls.1: src/ls
we have the laxer one:
man/ls.1: src/ls.c
and think of what happens if a user modifies, say, 'src/ls.c', and then
runs "make -j4" to rebuild everything. The make process will see that
it has to rebuild the man page 'man/ls.1' (because its prerequisite
'src/ls.c' has changed), but won't see that it has to rebuild 'src/ls'
*before* re-running 'help2man' to generate that man page; so, if
'man/ls.1' is rebuilt before 'src/ls' (which can happen with concurrent
make), our user will get either a build error (if 'src/ls' did non
exist) or, worse, a man page with an up-to-date timestamp but an
out-of-date content. And what's even worse in all of this is that
this problem will be present also for users who have perl installed:
this is not a "graceful degradation" at all!
In our situation, the best and simplest way to implement a graceful
degradation it to keep the correct dependencies for man pages (that
is, "man/ls.1: src/ls"), and if perl is not present, just generate
dummy man pages reporting that built-time issue and redirecting the
user back to either the info documentation or the '--help' output.
As a consequence of this change, we also stop distributing man pages,
since they will now be generated unconditionally.
* Makefile.am (do-not-require-help2man): Remove.
(dist-hook): Don't depend on it.
* man/local.mk: Remove an obsolete comment.
(EXTRA_DIST): Stop distributing generated man pages.
($(EXTRA_MANS)): This no longer needs to depend on $(all_programs).
(MAINTAINERCLEANFILES): $(ALL_MANS) Do not list it here, and ...
(CLEANFILES): ... list it here, instead.
(.x.1): Instead of warning if perl is missing, but then trying to run
'help2man' unconditionally, simply run ...
(run_help2man): ... the command referenced by this new variable, that
expands to a proper invocation of 'help2man' if perl is present, and
to an invocation of a shell script generating a dummy manpage if it
is not.
(EXTRA_DIST): Distribute that shell script.
* man/dummy-man: New shell script.
* NEWS (Build-related): Mention this.
Fixes coreutils http://bugs.gnu.org/12715.
Pádraig Brady [Wed, 24 Oct 2012 09:55:13 +0000 (10:55 +0100)]
pr: fix -n to pad consistently and not crash
* src/pr.c: Replace the code to truncate the most significant
digits of line numbers, with much simpler string manipulation
that supports arbitrary widths. Before this, specifying a
width >= 32 to -n would result in a divide by zero error.
Also remove the inconsistent padding with zeros and spaces, which
would result in zero padding for widths 12 and 15.
* tests/pr/pr-tests.pl: Added a test to ensure no zero padding,
and also a test for the divide by zero case.
* NEWS: Mentioned the fix
Reported by Ondrej Oprala
Pádraig Brady [Thu, 25 Oct 2012 09:59:23 +0000 (10:59 +0100)]
build: ensure factor links the iconv library
* src/local.mk (src_factor_LDADD): Append $(LIBICONV).
* crg.mk (sc_check-I18N-AUTHORS): A new syntax check rule
to ensure we add LIBICONV where appropriate.
* THANKS.in: Add the reporter.
Reported by Christian Jullien
Syntax check suggested by Jim Meyering
Pádraig Brady [Wed, 24 Oct 2012 22:18:23 +0000 (23:18 +0100)]
maint: refactor tests/misc/pr.pl into tests/pr/pr-tests.pl
* tests/misc/pr.pl: Refactor this test into ...
* tests/pr/pr-tests.pl: ... here.
* tests/local.mk: Remove the reference to the removed test
Improved by Jim Meyering
Pádraig Brady [Tue, 23 Oct 2012 10:59:04 +0000 (11:59 +0100)]
build: avoid warnings about unused variables and macros
src/factor.c (__GMP_DECLSPEC): Move back from longlong.h
to treat consistently with other stub macros.
(__GMP_GNUC_PREREQ): Reference to avoid -Wunused-macros warning.
(__GMP_DECLSPEC): Likewise.
(ASSERT): Likewise.
(__clz_tab): Likewise.
(factor_using_division): Mark a variable as unused.
(mulredc): Likewise.
(mulredc2): Likewise.
(divexact_21): Likewise.
Pádraig Brady [Tue, 23 Oct 2012 10:00:24 +0000 (11:00 +0100)]
build: avoid build failure on some sparc systems
* src/longlong.h: Restrict some sparc assembly variants
to sparc V9. This was seen to be an issue with newer
sparc systems with default gcc CPU options.
Jim Meyering [Tue, 16 Oct 2012 15:43:49 +0000 (17:43 +0200)]
cp: avoid data-corrupting free-memory-read
* src/extent-scan.c (extent_scan_read): Reset our last_ei
pointer whenever the parent buffer might have just been freed.
* tests/cp/fiemap-extent-FMR.sh: New test.
* tests/local.mk (all_tests): Add it.
* NEWS (Bug fixes): Mention it.
Reported by Mike Gerth in http://bugs.gnu.org/12656, and with
help from Alan Curry. Bug introduced in commit v8.10-60-g18f5a85.
Pádraig Brady [Mon, 8 Oct 2012 14:48:43 +0000 (15:48 +0100)]
factor: fix integer validation and GMP fallback
In the recent factor rewrite, the GMP code
wasn't actually used; just an error was printed
on integer overflow. While fixing that it was noticed
that correct input validation wasn't done in all cases
when falling back to the GMP code.
* src/factor.c (print_factors) Fallback to GMP on overflow.
(strto2uintmax): Scan the string for invalid characters,
so that case can be detected independently of overflow.
Return an error when an empty string is passed.
Also allow leading spaces and '+' in input numbers.
* tests/misc/factor.pl: Ensure the GMP code is exercised
when compiled in. Also add a test to verify leading
spaces and '+' are allowed.
Pádraig Brady [Mon, 8 Oct 2012 10:38:41 +0000 (11:38 +0100)]
build: support older GMP versions
The new factor code introduced usage of mpz_inits() and
mpz_clears(), which are only available since GMP >= 5,
and will result in a compile error when missing.
* m4/gmp.m4 (cu_GMP): Define HAVE_DECL_MPZ_INITS appropriately.
* src/factor (mpz_inits): New function, defined where missing.
(mpz_clears): Likewise.
Jim Meyering [Sun, 7 Oct 2012 08:37:25 +0000 (10:37 +0200)]
maint: avoid warning from gcc's -Wunused-macros
On some systems, -Wunused-macros would warn about two macros:
src/factor.c:148:0: warning: macro "__clz_tab" is not used
src/factor.c:126:0: warning: macro "UHWtype" is not used
* src/factor.c: Add a use to placate gcc.
Jim Meyering [Fri, 5 Oct 2012 20:35:27 +0000 (22:35 +0200)]
build: avoid link failure when strerror is replaced
When building the new make-prime-list program on a system for which
strerror is defined to rpl_strerror, we'd get a link failure.
The problem is that we're including <config.h> for some definitions,
but do not want the rpl_ ones, since this particular program must
not be linked against gnulib (aka libcoreutils.a). This did not
arise on Fedora 17 or 18, but did on Debian wheezy/sid.
* src/make-prime-list.c (strerror): #undef.
Build failure introduced by commit v8.19-152-gcf67e4c.
Jim Meyering [Tue, 18 Sep 2012 06:26:17 +0000 (08:26 +0200)]
factor: 25% speed-up, on output
* src/factor.c (print_factors_single): Use fputs and umaxtostr
rather than printf with "%ju". This reduced the time required
to compute and print the factors of the first 10^7 integers from
over 8 seconds to 5.75s. Run this command:
seq $((10**7)) | env time factor > /dev/null
Jim Meyering [Sun, 16 Sep 2012 20:31:04 +0000 (22:31 +0200)]
factor: merge with preexisting factor; integrate tests; avoid warnings
* src/factor.c: Renamed from factor-ng.c, with the following changes:
Adjust copyright header to be consistent with others.
Use xmalloc and xrealloc, to avoid segv upon OOM.
Switch back to using readtokens to handle input.
Diagnose invalid inputs.
s/fprintf+exit/error/
(print_factors): Add comments.
(strto2uintmax): Return strtol_error, not int.
(read_item): Remove, no longer used.
(main): Use atexit(close_stdout) so that we don't ignore failed write.
* cfg.mk: Exempt src/longlong.h from several tests.
Exempt run.sh from the test-list-consistency test.
Exempt make-prime-list.c from numerous tests, since we won't
be making it conform: it must not link with libcoreutils.a.
Exempt factor-ng.c from the no-upper-case error message test.
* AUTHORS (factor): Add Torbjörn and Niels.
* tests/local.mk (factor_tests): Encode the 37 tests.
($(factor_tests)): Rule to generate a test script for each test.
* tests/factor/run.sh: New script, marked as very expensive.
* .gitignore: Ignore new generated files.
* src/local.mk (src/primes.h): New rule.
(noinst_PROGRAMS): Add make-prime-list.
(noinst_HEADERS): Add longlong.h.
Remove all wheel-related rules and files.
* src/wheel-gen.pl: Remove file.
maint: mark set-but-not-used variables with ATTRIBUTE_UNUSED
* src/factor-ng.c (redcify, prime_p, isqrt2): Mark them, so we
don't have to disable -Wunused-but-set-variable.
maint: use __builtin_expect only if __GNUC__
* src/factor-ng.c (LIKELY, UNLIKELY) [__GNUC__]: Add #ifdef guard.
build: avoid warning about unused macro
* src/factor-ng.c (__GMP_DECLSPEC): Don't define here
* src/longlong.h (__GMP_DECLSPEC): Define if not already defined.
factor: new much-improved implementation; not yet integrated
* src/factor-ng.c: New file, from nt-factor.
* src/longlong.h: New file.
* NEWS (Improvements): Mention the upcoming improvements. Co-authored-by: Niels Möller
Jim Meyering [Sun, 23 Sep 2012 09:33:01 +0000 (11:33 +0200)]
maint: make-prime-list: do not ignore write failure
Even though this is just a helper program that is run solely to create
primes.h, it should not ignore a write failure. Normally we would
simply call atexit (close_stdout), but we cannot do that from this
helper program, since it must be built before the generated header,
primes.h. If we were to make the linking of make-prime-list depend
on libcoreutils.a, that would add all lib/*.o files to the list
of dependents of $(BUILT_HEADERS). Then, since there is currently no
provision to ensure that a file like lib/stdio.h (another built header)
is built before the first lib/*.o file that also includes <stdio.h>,
some lib/*.o files would be built before lib/stdio.h and some after.
The former would provoke link failures due to undefined rpl_* functions.
* src/make-prime-list.c: Include <errno.h>.
(fclose): Undef, so that a definition to rpl_fclose does not
cause a link failure.
(main): Per the above, in this exceptional case, we check for fclose
and ferror failure manually, and don't worry about the ferror-only
failure case in which errno may not be relevant.
* src/local.mk: Remove the above dependency.
A soon-to-be-added new program, make-prime-list, must not depend
on that, since it is used to create a BUILT_SOURCES file.
That dependency is already handled via the ..._LD_ADD variables,
and so that redundant dependency has so far been harmless.
Jim Meyering [Sun, 30 Sep 2012 12:28:47 +0000 (14:28 +0200)]
du: avoid abort on systems for which ->me_type is not malloc'd
On some systems (notably, BSD-based, like at least OpenBSD 4.9),
the me_type member does not come from the heap.
* src/du.c (fill_mount_table): Free the ->me_type member only
when it was malloc'd, i.e., when ->me_type_malloced is nonzero.
Bug introduced via commit v8.19-2-gcf7e1b5.
Reported as http://bugs.gnu.org/12542.
Ondrej Oprala [Tue, 7 Aug 2012 14:56:52 +0000 (16:56 +0200)]
cp: fix the --no-preserve=mode option
The --no-preserve=mode option did not do what its name implies:
it would mistakenly preserve permission mode bits.
* NEWS: Mention the fix.
* TODO: Remove an entry.
* src/copy.c (copy_reg): Add a condition to properly
handle the --no-preserve=mode option for files
(copy_internal): Add a condition to properly handle the
--no-preserve=mode option for directories.
* src/copy.h (struct cp_options): Add a new boolean.
* src/cp.c (cp_option_init,decode_preserve_arg): Set the
new boolean value according to specified options.
* src/install.c (struct cp_options): Initialize the new boolean.
* src/mv.c (struct cp_options): Initialize the new boolean.
* tests/cp/preserve-mode.sh: Add a new test.
* tests/cp/link-preserve.sh (-a --no-preserve=mode): Adjust the
expected perms: now, --no-preserve=mode overrides the --preserve=mode
that is inherent in -a, as it should.
* tests/local.mk: Add the new test to the list.
This was originally attempted in commit v8.12-117-g5a647a0,
but reverted before release because of the unreliability
of disabling core dumps using setrlimit() on Linux kernels.
This new version instead uses prctl() where available to
more reliably disable core dumps for the timeout process.
* m4/jm-macros.m4: Define HAVE_SETRLIMIT and HAVE_PRCTL.
* src/timeout.c (disable_core_dumps): A new function
that disables coredumps using prctl or setrlimit if available.
(main): If the child exited with a signal and we can
disable core dumps, then raise that signal to the timeout
process itself, so that callers may also see the signal status.
Also print a message indicating when the monitored command
dumped core, as that information is lost in the signal
propagation through timeout.
Pozsár Balázs [Mon, 24 Sep 2012 01:39:09 +0000 (02:39 +0100)]
dd: new option, status=none to suppress output statistics
* src/dd.c (STATUS_NONE): A new bitmask combining all STATUS_
options, thus used to suppress all informational output.
(struct symbol_value statuses): Expose the "none" option,
corresponding to the STATUS_NONE bitmask above.
(print_stats): Return early if STATUS_NONE is specified.
Also move the call to gethrxtime() down so that it's only
called when needed.
(usage): Describe the new options.
* doc/coreutils.texi (dd invocation): Likewise.
* NEWS: Mention the new feature.
* tests/dd/misc.sh: Ensure the new option works.
Jim Meyering [Wed, 19 Sep 2012 19:52:05 +0000 (21:52 +0200)]
tail,stat: improve support for vmhgfs
Teach tail -f that it must use polling on vmhgfs file systems, and
let stat -f --format=%T report the file system type name, "vmhgfs".
* src/stat.c (human_fstype): Add a case: vmhgfs, 0xbacbacbc, remote.
* NEWS (Improvements): Mention it.
* THANKS.in: Update.
Reported by Daniel Tschinder in http://bugs.gnu.org/12461.
Jim Meyering [Thu, 20 Sep 2012 15:47:25 +0000 (17:47 +0200)]
build: move non-recursive-gnulib-prefix-hack to gnulib; update
I've moved the non-recursive-gnulib-prefix-hack module to
gnulib, with two small improvements, so remove it from here
and update gnulib to the latest.
* gl/build-aux/prefix-gnulib-mk: Remove file.
* gl/m4/non-recursive-gnulib-prefix-hack.m4: Remove file.
* gl/modules/non-recursive-gnulib-prefix-hack: Remove file.
* gnulib: Update to latest.
Jim Meyering [Wed, 19 Sep 2012 07:35:12 +0000 (09:35 +0200)]
rm: be even more careful when using a replacement errno value
* src/remove.c (excise): The change in commit v8.19-107-gccbd3f3 made
the "rm -rf D" (for unreadable dir, D) diagnostic worse on Solaris 10:
-rm: cannot remove 'D': Permission denied
+rm: cannot remove 'D': File exists
That happened because unlinkat would fail with EEXIST there, given
an unreadable directory, which made the two tests, tests/rm/unread2
and tests/rm/unreadable fail. Accommodate the EEXIST case, too.
dd: remove references to the word BLOCKS from help and texi docs
The renaming from BLOCKS to N was done in v8.15-38-g140eca1,
and documentation for N was added again in v8.17-26-g4f2e9d5
without noticing that. Now, finally remove the word BLOCKS
from the documentation.
* src/dd.c (usage): Remove the word BLOCKS.
* doc/coreutils.texi (dd invocation): Likewise.
Improved by: Bernhard Voelker <mail@bernhard-voelker.de>
ls: fix coloring of dangling symlinks in default listing mode
When listing a directory containing dangling symlinks,
and not outputting a long format listing, and orphaned links
are set to no coloring in LS_COLORS, then the symlinks
would get no color rather than reverting to the standard
symlink color. The issue was introduced in v8.13-19-g84457c4
* src/ls.c (print_color_indicator): Use the standard method
to check if coloring is specified for orphaned symlinks.
The existing method would consider 'or=00' or 'or=0' as significant
in LS_COLORS. Even 'or=' was significant as in that case the
string='or=' and the length=0. Also apply the same change
for missing symlinks for consistency.
(gobble_file): Remove the simulation of linkok, which is only
tested in print_color_indicator() which now handles this directly
by keying on the LS_COLORS values correctly.
* tests/misc/ls-misc.pl: Add a test case.
* THANKS: Add the reporter.
* NEWS: Mention the fix. Reported-by: David Matei
seq: enable the fast integer printing code in more cases
* src/seq.c (main): Adjust the initial arbitrary precision
seq_fast enablement checks to be more maintainable, and
a little more general, by allowing single character
separators to use seq_fast.
Also check again after the number arguments are processed,
to see if we can still use seq_fast, which while not
allowing arbitarly large integers, it will handle
integers of the form 10E10 etc.
(seq_fast): Use a specified separator character,
rather than hardcoding '\n'.
Jim Meyering [Fri, 14 Sep 2012 17:38:33 +0000 (19:38 +0200)]
build: do not rely on automake's AM_TESTS_ENVIRONMENT
* tests/local.mk (TESTS_ENVIRONMENT): Rename from AM_TESTS_ENVIRONMENT,
since it is not honored in automake-1.11.3 after all.
This reverts commit v8.19-38-g34c9c8f. For now, I'll leave
the following commit that made bootstrap.conf require 1.11.2.
Prompted by a report of test failure from Pádraig Brady.
Jim Meyering [Thu, 13 Sep 2012 16:09:49 +0000 (18:09 +0200)]
seq: 70x faster for non-negative whole numbers and incr==1
Handle non-negative whole numbers robustly and efficiently when
the increment is 1 and when no format-changing option is specified.
On the correctness front, for very large numbers, seq now works fine:
The new code is much more efficient, too:
Old vs new: 55.81s vs 0.82s
$ env time --f=%e seq $((10**8)) > /dev/null
55.81
$ env time --f=%e src/seq $((10**8)) > /dev/null
0.82
* seq.c (incr): New function, inspired by the one in cat.c.
(cmp, seq_fast): New functions, inspired by code in nt-factor
by Torbjörn Granlund and Niels Möller.
(trim_leading_zeros): New function, without which cmp would malfunction.
(all_digits_p): New function.
(main): Hoist the format_str-vs-equal_width check to precede first
treatment of operands, and insert code to call seq_fast when possible.
* NEWS (Bug fixes): Mention the correctness fix.
(Improvements): Mention the speed-up.
* tests/misc/seq.pl: Exercise the new code.
Improved by: Bernhard Voelker.
http://thread.gmane.org/gmane.comp.gnu.coreutils.general/3340
Jim Meyering [Fri, 14 Sep 2012 07:25:33 +0000 (09:25 +0200)]
tests: reenable "make -C tests ..." commands
The README-documented way to run individual tests was invalidated
by the conversion of tests/ to non-recursive make. Add a GNUmakefile
shim to reenable that usage.
* tests/GNUmakefile: New file, so that "make -C tests ..." works
like it did before the conversion of tests/ to non-recursive build.
Reported by Bernhard Voelker.
* Makefile.am (EXTRA_DIST): Add it.
* cfg.mk (sc_prohibit_tab_based_indentation): Also exempt any
GNUmakefile from this syntax-check.
build: fixup: correctly recognize if perl is missing
* configure.ac: Here, by adding a missing '*' to the wildcard in
a 'case' construct over the contents of $PERL. Introduced in
commit v8.19-41-g00f5ba1.
Jim Meyering [Sun, 9 Sep 2012 06:21:16 +0000 (08:21 +0200)]
build: build lib/ using non-recursive make
* bootstrap.conf (gnulib_modules): Use the new module.
(bootstrap_post_import_hook): Invoke prefix-gnulib-mk.
* configure.ac (AC_CONFIG_FILES): Remove lib/Makefile.
* lib/Makefile.am: Renamed...
* lib/local.mk: ...to this.
* src/local.mk (CLEANFILES): Append, don't set.
(noinst_LIBRARIES): Likewise.
(AM_CPPFLAGS): Don't set this here.
* Makefile.am (AM_CPPFLAGS): Define here instead.
(noinst_LIBRARIES, CLEANFILES, MOSTLYCLEANDIRS, MOSTLYCLEANFILES):
Initialize here, so we can append to them from each included local.mk
(SUBDIRS): Remove "lib".
Tell the prefix-adding script not to add a prefix when the word it's
prefixing is "@ALLOCA@". That is fine for most cases, but what about
when the expansion of @ALLOCA@ is nonempty?
* build-aux/prefix-gnulib-mk (prefix_word): Exclude @ALLOCA@.
* gl/m4/non-recursive-gnulib-prefix-hack.m4: Prefix non-empty
$ALLOCA with "lib/". FIXME: I'm not sure this is required,
now that we...
Use AC_CONFIG_LIBOBJ_DIR([lib]).
Without using AC_CONFIG_LIBOBJ_DIR([lib]), automake (not autoconf)
would complain of failure to find aclocal.c, due to the use of
AC_LIBSOURCES(alloca.c).
Jim Meyering [Mon, 10 Sep 2012 14:38:03 +0000 (16:38 +0200)]
build: new module to convert lib/ to non-recursive make
* gl/modules/non-recursive-gnulib-prefix-hack: New module.
* gl/m4/non-recursive-gnulib-prefix-hack.m4:
(gl_NON_RECURSIVE_GNULIB_PREFIX_HACK): This is the snippet
that this module inserts near the end of configure.
* gl/build-aux/prefix-gnulib-mk: New script, from bison.
Changes from the code in bison:
(prefix_assignment): Split a long line.
(prefix): Add trailing slashes to avoid a single false match.
Prefix imaxtostr.c and the other *tostr.c file names manually.
Also, use $prefix in place of hard-coded "lib/".
* gl/lib/mbsalign.h: Add MBA_UNIBYTE_ONLY (to allow
faster processing). Also add MBA_NO_LEFT_PAD, MBA_NO_RIGHT_PAD
to give greater control of padding, useful with the first
or last fields on a line.
* gl/lib/mbsalign.c (mbsalign): Implement the new flags.
* gl/tests/test-mbsalign.c (main): Test combinations
of the new flags.
Jim Meyering [Sat, 8 Sep 2012 19:17:32 +0000 (21:17 +0200)]
maint: allow per-directory CFLAGS settings with non-recursive make
* src/local.mk (AM_CFLAGS): Don't use $(WARN_CFLAGS) here.
* cfg.mk (src_CFLAGS, lib_CFLAGS, gnulib-tests_CFLAGS): Define here
instead.
(AM_CFLAGS): Augment using the above.
* configure.ac: Note that the configure-time option,
--enable-gcc-warnings now functions only when using GNU make.
Well, currently it does still work in gnulib-tests, but that should
soon be fixed. Improved-by: Stefano Lattarini
Jim Meyering [Sun, 9 Sep 2012 17:27:25 +0000 (19:27 +0200)]
build: do not require help2man at build-from-tarball time
But do retain full dependencies when building from a git clone.
We do this by converting the full dependency (of the .1 file on
the binary we run with --help) into a dependency on the .c file.
* Makefile.am (do-not-require-help2man): New rule.
(dist-hook): depend on it.
Jim Meyering [Sat, 8 Sep 2012 13:11:20 +0000 (15:11 +0200)]
maint: move the last coreutils-specific files from lib/ to gl/
Convert the few remaining coreutils-specific files in lib/ to
gnulib-style modules under gl/, removing their corresponding .m4
files, since the information recorded in those files is better
stored in module-description file in gl/modules/.
* bootstrap.conf (gnulib_modules): Add new modules:
fd-reopen, buffer-lcm, xfts, strnumcmp.
* gl/lib/buffer-lcm.c: Renamed from the file in lib/.
* gl/lib/buffer-lcm.h: Likewise.
* gl/lib/fd-reopen.c: Likewise.
* gl/lib/fd-reopen.h: Likewise.
* gl/lib/strintcmp.c: Likewise.
* gl/lib/strnumcmp-in.h: Likewise.
* gl/lib/strnumcmp.c: Likewise.
* gl/lib/strnumcmp.h: Likewise.
* gl/lib/xfts.c: Likewise.
* gl/lib/xfts.h: Likewise.
* gl/modules/buffer-lcm: New module-description file.
* gl/modules/fd-reopen: Likewise.
* gl/modules/strnumcmp: Likewise.
* gl/modules/xfts: Likewise.
* m4/fd-reopen.m4: Remove, no longer needed.
* m4/strnumcmp.m4: Likewise.
* m4/xfts.m4: Likewise.
* m4/prereq.m4: Do not AC_REQUIRE the m4 functions from
our just-removed m4/*.m4 files.
Jim Meyering [Sat, 8 Sep 2012 14:01:47 +0000 (16:01 +0200)]
maint: move gl/m4/root-dev-ino.m4 contents into its module definition
We can get the same effect using the modules file.
* gl/m4/root-dev-ino.m4: Remove file.
* gl/modules/root-dev-ino (Depends-on): Add lstat.
(Files): Remove m4/root-dev-ino.m4.
(Makefile.am) [lib_SOURCES]: Add root-dev-ino.c and root-dev-ino.h.
(configure.ac): Remove reference to gl_ROOT_DEV_INO.
The multiple-precision factoring code (with HAVE_GMP) was copied from
a now-obsolete version of GMP that did not pass proper arguments to
the mpz_probab_prime_p function. It makes that code perform no more
than 3 Miller-Rabin tests only, which is not sufficient.
A Miller-Rabin test will detect composites with at least a probability
of 3/4. For a uniform random composite, the probability will actually
be much higher.
Or put another way, of the N-3 possible Miller-Rabin tests for checking
the composite N, there is no number N for which more than (N-3)/4 of the
tests will fail to detect the number as a composite. For most numbers N
the number of "false witnesses" will be much, much lower.
Problem numbers are of the form N=pq, p,q prime and (p-1)/(q-1) = s,
where s is a small integer. (There are other problem forms too,
involving 3 or more prime factors.) When s = 2, we get the 3/4 factor.
It is easy to find numbers of that form that cause coreutils' factor to
fail:
There are 9008992 composites of the form with s=2 below 2^64. With 3
Miller-Rabin tests, one would expect about 9008992/64 = 140766 to be
invalidly recognized as primes in that range.
* src/factor.c (MR_REPS): Define to 25.
(factor_using_pollard_rho): Use MR_REPS, not 3.
(print_factors_multi): Likewise.
* THANKS.in: Remove my name, now that it will be automatically
included in the generated THANKS file.
Jim Meyering [Thu, 6 Sep 2012 10:00:16 +0000 (12:00 +0200)]
tests: improve checks for setuidgid-using root-only tests
* init.cfg (setuidgid_has_perm_): New function.
(require_root_): Use it. Improved-by: Bernhard Voelker
* NEWS (Build-related): Mention the improvement.
Eric Blake [Thu, 6 Sep 2012 14:20:17 +0000 (08:20 -0600)]
build: default to --enable-gcc-warnings for git tree
Anyone developing on coreutils can be assumed to have a new enough
environment, such that enabling gcc warnings by default will be
useful. Tarballs still default to no warnings, and the defaults
can still be overridden with --disable-gcc-warnings.
* configure.ac (gl_gcc_warnings): Set default based on environment.
Suggested by Bernhard Voelker.
Jim Meyering [Wed, 5 Sep 2012 20:33:44 +0000 (22:33 +0200)]
tail,stat: add support for VZFS
* src/stat.c (human_fstype): Add a case: vzfs, 0x565A4653 (local).
Reported by Jens Rosenboom in http://bugs.gnu.org/12356
* NEWS (Improvement): Mention it.
Jim Meyering [Wed, 5 Sep 2012 14:48:50 +0000 (16:48 +0200)]
rm: be more careful when using a replacement errno value
* src/remove.c (excise): Tighten the test for when we defer to an
old errno value: instead of relying solely on an FTS_DNR (unreadable
directory) failure, also test current and replacement errno values.
This change would also have solved the problem addressed by commit v8.19-106-g57dd067. For more info, see http://bugs.gnu.org/12339#113
Jim Meyering [Tue, 4 Sep 2012 12:40:11 +0000 (14:40 +0200)]
rm: avoid bogus diagnostic for a slash-decorated symlink-to-dir
These commands would evoke an invalid diagnostic:
$ mkdir d && ln -s d s && env rm -r s/
rm: cannot remove 's': Too many levels of symbolic links
remove.c was stripping trailing slashes from "s/" before passing
the name to "rm". But a trailing slash may change the semantics,
and thus should not be stripped.
* src/remove.c (rm_fts): Do not strip trailing slashes.
* tests/rm/v-slash.sh: Adapt to new expected output.
* gnulib: Update to latest, for an improved fts.c that merely
normalizes trailing slashes.
Reported by Paul Eggert in discussion of http://bugs.gnu.org/12339
Jim Meyering [Tue, 4 Sep 2012 10:12:19 +0000 (12:12 +0200)]
tests: depend on the programs we're testing
* tests/local.mk ($(TEST_LOGS)): Depend on $(PROGRAMS), so that
tests are rerun when any program is rebuilt. Technically, we could
specify precisely which few programs are dependents of each test,
but that can come later, if deemed worth the trouble and maintenance
burden. Also, there is the issue of the primary program(s) being
tested (i.e., those itemized via print_ver_) versus those that are
tested incidentally: for example, nearly every test exercises "rm"
when its clean-up code removes files.
* cfg.mk: Don't work by trying to parse the (now gone) file
'tests/Makefile.am'; rather, use the contents of the make variable,
$(all_root_tests), introduced few commits ago.
Fix a few unrelated cosmetic issues while at it.
build: don't use recursive make for tests/ subdirectory
* Makefile.am (SUBDIRS): Remove 'tests'.
(include): The '$(top_srcdir)/tests/local.mk' file.
(check-root): Remove this convenience target, it's no longer needed
now that the "real" check-root target once in 'tests/Makefile' will
land in the top-level makefile.
* configure.ac (AC_CONFIG_FILES): Remove 'tests/Makefile'.
* tests/Makefile.am: Rename ...
* tests/local.mk: ... like this, with a lot of adjustments.
* tests/init.cfg: Move ...
* init.cfg: ... here. This is necessary, for a limitation of the
gnulib-provided 'tests/init.sh', which unconditionally look for
'init.cfg' in the $(srcdir) directory.
* tests/*/*.sh: Adjust: expect init.sh to be in '$srcdir/tests',
not in '$srcdir', and extend $PATH with './src', not with '../src'.
* tests/Coreutils.pm: Adjust similarly.
* tests/pr/pr-tests.pl ($pfx): Likewise.
maint: list of names of built programs available in the Makefiles
This is just a preparatory refactoring in view of future changes.
* configure.ac (AC_SUBST): New 'built_programs'.
* tests/Makefile.am (AM_TESTS_ENVIRONMENT): Simply define the exported
variable 'built_programs' to the expansion of the '$(built_programs)'
AC_SUBST'd make variable.
(.built-programs): Remove this now-unneeded convenience target.
(CLEANFILES, check_DATA): Delete, no longer needed.
* cfg.mk (sc_no_exec_perl_coreutils): This. Our new testsuite
layout (perl tests having '.pl' suffix, shell tests having '.sh'
suffix) makes it basically impossible to run into the issue this
check guarded against.
* tests/Makefile.am (root-hint): Here. The interested user can see
the reasons why some tests are skipped by looking at the messages
they display on the console; here's an excerpt:
...
PASS: misc/id-groups.sh
id-setgid.sh: skipped test: must be run as root
SKIP: misc/id-setgid.sh
PASS: misc/md5sum.pl
...
PASS: df/total-verify.sh
2g.sh: skipped test: very expensive: disabled by default
SKIP: du/2g.sh
...
Clear enough, and more specific and precise that a generic "some tests
might need to be run as root" message. And if that user is interested
in making those tests run anyway, he'll just take a look to the README
files to look for info. So there's no reason to pollute the stdout
with another "hint" that is subsumed by those messages, and that might
go unnoticed anyway.
Moreover, and possibly more importantly, that hint wasn't being
displayed anyway, even before this change! That's because the
'root-hint' target was listed as prerequisite for the 'check-recursive'
target, which however was not a dependency of the 'check' target in
'tests/Makefile.am', because that file contains no $(SUBDIRS)
definition.
tests: use suffix, not exec-bit, to tell which files are tests
* tests/Makefile.am (vc_exe_in_TESTS): Adjust to look, in the 'tests/'
subdirectory, for files that have one of the extensions listed in
$(TEST_EXTENSIONS), rather than for executable files.
maint: make vc_exe_in_TESTS also work in VPATH builds
* tests/Makefile.am (vc_exe_in_TESTS): It is easy to adjust this
recipe to also work in VPATH setups, also thanks to modifications
done by previous changes.
maint: avoid parsing of Makefile.am from vc_exe_in_TESTS
* tests/Makefile.am (TESTS): Rename ...
(all_tests): ... like this, so that we'll still be able to know the
complete list of our tests even if the user overrides TESTS from the
command line (which he's allowed to do by the test harness API).
(root_tests): Rename ...
(all_root_tests): ... like this, for similar reasons.
(TESTS, root_tests): Redefine their defaults to to $(all_tests) and
$(all_root_tests) respectively.
(vc_exe_in_TESTS): It can now safely use $(all_tests) to get the
complete list of test cases according to the Makefile, instead of
having to resort to "parsing" of Makefile.am.
(EXTRA_DIST): Use $(all_tests), not $(TESTS).
(v_, w_): Delete, no longer needed.
build: use 'check-local' to extend the 'check' target
* tests/Makefile.am (check-local): Here, by making this depend
on 'vc_exe_in_TESTS' ...
(check): ... rather than making this depend on them. While the old
usage worked, it relied on an implementation detail rather than on
documented behavior.
* src/local.mk (check-local): Similarly, make this depend on
'check-README' and 'check-duplicate-no-install' ...
(check): ... rather than on this.
scripts: autotools-install: allow user-specified make program
* scripts/autotools-install: Honor $MAKE. This might be useful
on systems where the make implementation available in $PATH
by default is limited (Solaris) or broken (HP-UX).
Jim Meyering [Sun, 2 Sep 2012 11:46:17 +0000 (13:46 +0200)]
maint: disable sc_proper_name_utf8_requires_ICONV test
It would still pass, but would print many diagnostics like this:
Can't open src/Makefile.am: No such file or directory.
* cfg.mk (local-checks-to-skip): Temporarily disable a test.
This test will need to be adapted to work with a non-recursive
build set-up, in which there is no Makefile.am alongside each program.
Reported by Bernhard Voelker.
Jim Meyering [Sun, 2 Sep 2012 11:42:39 +0000 (13:42 +0200)]
maint: accommodate upcoming, expanded list of warnings from gnulib
* configure.ac: Disable a new gcc warning, -Wsuggest-attribute=format,
since it triggers on copy.c (which I'm not inclined to adjust) and
factor.c's use of vfprintf which would appear to require a change
to stdio.h.
projects / thirdparty / coreutils.git / log
