java: check_JAVA does not cause compilation by "make all" anymore
Fixes automake bug#8234.
* automake.in (handle_java): Make stamp of class files built from
java sources in $(check_JAVA) a dependency of `check' target, not
`all' target.
* tests/java-check.test: New test.
* tests/Makefile.am (TESTS): Update.
* NEWS: Update.
* THANKS: Update.
coverage: more on java support: EXTRA_ and noinst_ prefixes
* tests/java-extra.test: New test, checking support for the
prefix `EXTRA_' with the JAVA primary.
* tests/java-noinst.test: New test, checking support for the
prefix `noinst_' with the JAVA primary.
* tests/Makefile.am (TESTS): Update.
* lib/depcomp [makedepend]: Remove any VPATH prefix from the
object file name, so a rebuild doesn't attempt to update the
.Po files in the source tree.
* tests/depcomp9.test: New test.
* tests/Makefile.am (TESTS): Update.
* NEWS: Update.
Create subdirs for generated sources even when not dep tracking.
* automake.in (handle_single_transform): If $object is derived
and lands in subdir, be sure to output a dirstamp dependency.
* tests/yacc5.test: Avoid falsely matching the dirstamp
dependency when grepping for a rule.
* tests/lex-subobj-nodep.test: New test.
* tests/Makefile.am (TESTS): Update.
* THANKS: Update.
Report by Ignacy Gawedzki.
* doc/automake.texi (@menu): Rename title for entry 'configure'
from "Scanning configure.ac or configure.in" to the more precise
"Scanning configure.ac, using aclocal".
(@detailmenu): Adjust.
(@node configure): Adjust, and extend @cindex calls accordingly.
* THANKS: Update.
* tests/aclocal6.test: Move checks related to "maintainer-clean"
functionalities into ...
* tests/maintclean-vpath.test: ... this new test.
* tests/maintclean.test: Update heading comments. Extend to also
test subdirs. Remove useless disabling of YACC. Fix m4 quoting
in configure.in. Add a trailing `:' command.
* tests/Makefile.am (TESTS): Update.
yacc: extension of headers modelled after extension of sources
With this change, if '-d' is in *YFLAGS, a yacc input file named
foo.y++ will cause a foo.h++ header to be generated, instead of a
foo.h header. Similarly for foo.ypp, foo.yxx and foo.yy.
This way, the name of the files generated by an automake-created
`ylwrap' invocation should be consistent with those generated by
a `bison -o' call.
Related to automake bug#7648 and PR automake/491.
* lib/am/yacc.am (am__yacc_c2h): New internal variable.
(?GENERIC?%EXT%%DERIVED-EXT%, ?!GENERIC?%OBJ%): Get the name of
the header dynamically at make runtime, so that its extension is
modelled after the extension of the source.
* automake.in (lang_yacc_target_hook): Adjust the calculation of
`$header' accordingly.
* tests/yacc-cxx.test: New test.
* tests/yacc-d-cxx.test: Likewise.
* tests/yacc-weirdnames.test: Likewise.
* tests/yacc-basic.test: Update comments.
* tests/yacc-d-basic.test: Likewise.
* tests/yaccpp.test: Updated and extended.
* tests/Makefile.am (TESTS): Update.
* tests/deprecated-acinit.test: New test, check that automake
and autoconf complain about an old-style AC_INIT call used with
a new-style AM_AUTOMAKE_INIT call.
* tests/Makefile.am (TESTS): Update.
maintcheck: look for problematic names of testcases
The configure.in stub created by default by `tests/defs' obtains
the first argument of AC_INIT from the test name, and this can
cause some supported autoconf versions to fail with a spurious
error if that test name contains the name of an m4 or m4sugar
builtin or macro (e.g., `defn' or `m4_undefine').
This change add a maintainer check that warns about test names
which are possibly problematic in this regard.
* Makefile.am (sc_test_names): New maintainer-check target.
(syntax_check_rules): Add it.
(m4_builtins): New helper variable.
(TESTS): Updated according to the following renamings.
* tests/include.test: Renamed ...
* tests/hdr-vars-defined-once.test: ... to this.
* tests/sinclude.test: Renamed ...
* tests/m4-inclusion.test: ... to this, and simplified
accordingly.
* tests/include2.test: Renamed ...
* tests/dist-included-parent-dir.test: ... to this, for
consistency.
* tests/alloca.test: Make grepping of automake stderr stricter,
add a trailing `:' command; also, add AC_PROG_CC to configure.in,
and create a dummy alloca.c file, to ensure that we fail for the
proper reason.
* tests/alloca2.test: Likewise. Also, look for LT_INIT, not
AC_PROG_LIBTOOL, in the error message (bug introduced with commit v1.11-315-gd51e7b7 "libtool: suggest LT_INIT if LTLIBRARIES
primary is used").
* tests/comments-in-var-defn.test: The configure.in stub created
by default, which has the AC_INIT first argument obtained by the
test name, causes autoconf 2.62 to fail with a spurious error
message like: "configure.in:1: error: defn: undefined macro:".
Thus, to prevent this, the test is renamed to ...
* tests/comments-in-var-def.test: ... this.
* tests/Makefile.am (TESTS): Updated.
tests: extend checks on then 'unindent' shell function
* tests/self-check-unindent.test: Also check that multiple
instances of 'unindent' can run in parallel (this was not
the case when that function used temporary files).
Peter Rosin [Tue, 1 Mar 2011 07:25:24 +0000 (08:25 +0100)]
test defs: unindent without temporary file
* tests/defs.in (commented_sed_unindent_prog): Commented Sed program
that strips the "proper" amount of leading whitespace.
(unindent): Lazily strip comments from the above program and use it
to unindent without using a temporary file.
libtool: suggest LT_INIT if LTLIBRARIES primary is used
When the LTLIBRARIES primary was used, but $(LIBTOOL) wasn't
defined, automake suggested to add a call to AC_PROG_LIBTOOL
in configure.ac. But that macro is deprecated since Libtool
version 1.9b (2004-08-29), in favor of the newer LT_INIT. So
suggest the use of this latter macro instead.
* lib/Automake/Variable.pm (%_am_macro_for_var): Pair 'LIBTOOL'
with 'LT_INIT', not with 'AC_PROG_LIBTOOL'.
* libtool4.test: Adjust and extend. Also, add a call to macro
AC_PROG_CC in configure.in, to help ensuring that automake does
not fail for the wrong reasons.
* ltinit.test: New test, ensure that automake's libtool support
works with LT_INIT-based interface.
tests: add testcases sanity-checking the testsuite
Helper subroutines, variables and other pieces of code defined
in the `tests/defs' and used by many testcases are non-obvious,
and tricky to get to work portably; but until now, they weren't
tested at all in a clear and self-contained way.
This change should remedy to the situation.
* tests/self-check-cleanup.test: New test, check removal of
temporary test working directory by `./defs'.
* tests/self-check-dir.test: New test, check that tests using
`./defs' create a proper temporary directory, and run in it.
* tests/self-check-exit.test: New test, check that, in case of
failing commands, the correct exit status is passed to the exit
trap installed by the `./defs' script.
* tests/self-check-is_newest.test: New test, checking the
`is_newest' subroutine.
* tests/self-check-me.test: New test, checking that $me gets
defined automatically by `tests/defs' if not set, and that it
can be overridden from either the shell or the environment.
* tests/self-check-sanity.test: New test, check that the sanity
checks performed by the `tests/defs' script works correctly.
* tests/self-check-unindent.test: New test, checking the
`unindent' subroutine.
* tests/Makefile.am (TESTS): Update.
tests: tempdirs with restrictive permissions are cleaned correctly
Before this change, the removal of a temporary test directory
containing subdirectories with restrictive permissions (such as
'r--r--r--') could fail.
* tests/defs: Ensure that all the subdirectories of a temporary
test directory have the 'read', 'write' and 'execute' bits set,
before trying to remove it with `rm -rf'.
* tests/Makefile.am (clean-local-check): Likewise.
After the split of `instspc.test' into various generated tests,
the running time of the testsuite has noticeably increased, since
all these new generated tests must run aclocal, autoconf and
automake, whereas previously they were run only once (at the
beginning of `instspc.test'). But luckily, since the new tests
share the same input files for the autotools, this situation can
be easily worked around (at the expenses of a slight increase of
complexity for the testsuite scaffolding).
* tests/instspc-data.test: New helper test, properly calling
the `instspc-tests.sh' script to generate input data for the
others `instspc-*.test' tests.
* tests/Makefile.am (TESTS): Add `instspc-data.test'.
($(instspc_tests:.test=.log)): Depend on its log file.
(instspc-data.log): Depend on `instspc-tests.sh'.
* tests/instspc-tests.sh: Recognize new action `generate-data',
and use it to create hand-written and autotools-generated static
files shared by all the `instspc-*.test' tests.
When sourced by the `instspc-*.test' tests, use those previously
created files instead of recreating them from scratch.
(unindent, create_input_data): New subroutines.
Some other related changes and refactorings.
* tests/python-vars.test: Ignore cached values from config.site
by exporting CONFIG_SITE=/dev/null. Be laxer in matching the
expected values of output variables `pythondir' and `pyexecdir',
since they can change quite unpredictably among different python
installations. Also, avoid "hyping" debugging output, thus
offering smaller trace output and more informative diff.
* tests/python-virtualenv.test: Require python, since we call it
even after the virtualenv has been deactivated. Ignore cached
values from config.site by exporting CONFIG_SITE=/dev/null.
tests: tweak few tests on simple and parallel test drivers
* tests/check-exported-srcdir.test: Improve heading comments.
* tests/check-tests-in-builddir.test: Likewise. Also, unset the
`FOO_EXIT_STATUS' variable, so that any pre-existing value in the
environment won't risk to interfere with the test.
python: extend and improve tests, fix minor glitches
* m4/python.m4 (AM_PATH_PYTHON): Ensure the "checking ..." messages
from configure are always complete, even in case of failure. Tiny
cosmetic improvement in info/error messages.
* tests/python.test: Also check that automake complains if the
PYTHON primary is used but the `py-compile' script is not present.
Make grepping of generated Makefile.in laxer w.r.t. whitespace.
Add trailing `:' command.
* tests/python2.test: Remove repeated calls aclocal: they are
useless because configure.in is never modified. Make grepping
of automake stderr more comprehensive. Remove the pre-existing
`py-compile' file before trying to install it with `--add-missing'.
Add trailing `:' command.
* tests/python3.test: Add trailing `:' command.
* tests/python11.test: Likewise.
* tests/python4.test: Likewise. Also, try to pass PYTHON config
variable to configure from the environment, rather than only from
the command line.
* tests/python5.test: Ensure that the "checking ..." messages from
configure are always complete. Use proper m4 quoting. Add a
trailing `:' command.
* tests/python6.test: Simplify test logic, by checking for files
created by configure rather then grepping its output.
* tests/python7.test: Likewise.
* tests/python8.test: Also check that `$PYTHON' is meaningfully
set in the ACTION-IF-TRUE argument of AM_PATH_PYTHON.
* tests/python9.test: Likewise.
* tests/python10.test: Add trailing `:' command.
(Makefile.am): Declare `disttest' target as `.PHONY', and add
an `ls -l' to its recipe, for debugging.
* tests/nobase-python.test: In testing "make uninstall" and
"make install" results, prefer idioms that make verbose logs
more helpful. Remove a couple of lines of dead code. Add a
trailing `:' command.
* tests/python5b.test: New test, checking that configure performs
the check on the python version even when the choice of the python
interpreter is forced by the user.
* tests/python-dist.test: New test, checking the distribution of
*_PYTHON files.
* tests/python-vars.test: New test, checking that AM_PATH_PYTHON
correctly set all the output variables advertised in the manual.
* tests/python-virtualenv.test: New test, checking that python
support offered by automake works well with "virtual python
environments" created by the `virtualenv' program.
* tests/instdir-prog.test: Also check `$(pyexecdir)'. Existing
checks made slightly stricter.
* tests/instdir-prog.test: Also check `$(pyexecdir)'. Existing
* tests/instdir-ltlib.test: Likewise.
* tests/Makefile.am (TESTS): Update.
We don't want YACC and LEX defined as make variables, otherwise
the values determined at configure time will override those from
the environment, even in the make-spawned testcases. For example,
before this change, with the following usage:
$ ./configure YACC=yacc
$ export YACC='bison -y'
$ make check
the testsuite would have ended up, very counterintuitively, with
YACC defined to 'yacc' in the testcases' environment.
* configure.ac: Call `AM_SUBST_NOTMAKE' on YACC and LEX.
tests: remove redundancy from silent lex/yacc tests
* tests/silent-yacc-gcc.test, tests/silent-yacc-generic.test: Merge
these two testcases into ...
* tests/silent-yacc.test: ... this new one, which doesn't fiddle
which dependency tracking, as that shouldn't impact on generation
of yacc-derived C source and header files (and, for what concerns
compilation of C files, is already tested in other testcases).
* tests/silent-lex-gcc.test, tests/silent-lex-generic.test: Merge
these two testcases into ...
* tests/silent-lex.test: ... this new test, for similar reasons.
* tests/Makefile.am (TESTS): Update.
tests init: better messages for 'yacc' and 'lex' requirements
* tests/defs.in: Give better diagnostic messages when a test must
be skipped to the unavailability of yacc or lex program. Also,
improve syncing between code for requiring yacc and lex.
configure: look for a lex program to be used by the testsuite
This will allow the testcases requiring a 'lex' program to run also
with vendor/legacy lex implementations, not only with 'flex'.
* configure.ac: Look for a lex program, using AC_CHECK_PROGS.
* tests/defs.in: New required entry 'lex'.
($LEX): Let the user override the lex program to be used by the
testsuite.
* tests/cond35.test ($required): Require 'lex', not 'flex'.
* tests/cond36.test: Likewise.
* tests/lexv3.test: Likewise.
* tests/lexv3.test: Likewise.
* tests/silent-lex-gcc.test: Likewise.
* tests/silent-lex-generic.test: Likewise.
* tests/silent-many-gcc.test: Likewise.
* tests/silent-many-generic.test:likewise.
* tests/lexvpath.test: Likewise, and fix typo in comments.
* doc/automake.texi (Simple Tests): Document that automatic
recognition of a capable terminal attached to stdout can fail
with some make implementation when running in parallel mode,
thus causing colored test output not to be automatically
activated when it should.
Ralf Wildenhues [Sun, 23 Jan 2011 10:26:50 +0000 (11:26 +0100)]
tests: avoid instspc.test failures due to make's whitespace trimming
* tests/instspc.test: Prepend './' when passing the test
characters, to avoid leading whitespace characters to be trimmed
from macros set from environment variables. Fixes testsuite
failures with HP-UX, IRIX, and Tru64/OSF make.
coverage: test semantics of "dummy" per-target flags
* tests/specflg-dummy.test: New test, ensuring that even "dummy"
per-target flags triggers the use of renamed objects.
* tests/Makefile.am (TESTS): Update.
* tests/defs.in ($IFS): Define to <space>, <tab>, <newline>.
($sp): New variable, holding a single whitespace character.
($tab): New variable, holding a tabulation character.
($nl): New variable, holding a newline character.
configure: look for a yacc program to be used by the testsuite
Instead of forcing the user to manually export 'YACC' in the
testsuite to use a non-bison yacc, we now look for a yacc program
at configure time, and use that as the default in the testsuite.
* configure.ac: Look for a yacc program, using AC_CHECK_PROGS.
* tests/defs.in: Updated to use the value of $YACC precomputed by
configure, unless the user overrides that in the environment.
Ralf Wildenhues [Sun, 23 Jan 2011 09:57:21 +0000 (10:57 +0100)]
tests: avoid instspc* failures due to make's whitespace trimming
* tests/instspc-tests.sh: Prepend './' when passing the test
characters, to avoid leading whitespace characters to be trimmed
from macros set from environment variables. Fixes testsuite
failures with HP-UX and Tru64/OSF make.
Ralf Wildenhues [Sat, 22 Jan 2011 12:11:39 +0000 (13:11 +0100)]
tests: fix VPATH auto-expansion workarounds.
* tests/suffix10.test, tests/suffix11.test, tests/suffix12.test,
tests/suffix3.test, tests/suffix5.test, tests/suffix8.test:
Ensure $< is not surrounded by white space, to prevent Solaris
make from applying automatic VPATH text expansion.
tests: more coverage on yacc/lex silent-rules, plus minor cleanups
* tests/silent-yacc-gcc.test: Add sanity checks verifying that the
generated Makefile.in files really contains the non-generic rules
we expect. Do not redundantly manually remove files we know to be
already removed "make maintainer-clean".
(Makefile.am): Ensure we cover also non-generic yacc rules, by
setting target-specific YFLAGS.
(sub/Makefile.am): Likewise.
* tests/silent-yacc-generic.test: Likewise.
* tests/silent-lex-gcc.test: Likewise, but with LFLAGS instead of
YFLAGS.
* tests/silent-lex-generic.test: Likewise.
* tests/silent-many-gcc.test: Likewise, but with both LFLAGS and
YFLAGS. Also ...
(do_and_check_verbose_build): Remove redundant blank line.
* tests/silent-many-generic.test: Likewise.
* tests/silent-yacc-headers.test: New test.
* tests/Makefile.am (TESTS): Update.
Ralf Wildenhues [Sat, 22 Jan 2011 17:20:41 +0000 (18:20 +0100)]
tests: fix VPATH auto-expansion workarounds.
* tests/parallel-tests8.test, tests/suffix13.test:
Ensure $< is not surrounded by white space, to prevent Solaris
make from applying automatic VPATH text expansion.
Ralf Wildenhues [Sat, 22 Jan 2011 12:46:27 +0000 (13:46 +0100)]
tests: avoid fn99*.test failures due to buggy AIX 5.3 cp -R.
* tests/fn99.test, tests/fn99subdir.test: Skip if an initial
`cp -R' of the subdir tree already fails; AIX 5.3 cp messes
up its internal memory when copying this tree.
tests: cover yacc target-specific flags, and `-v' flag handling
* tests/yacc-basic.test: Also try to build a parser having `-v'
as target-specific flags. Add a couple of `ls -l' commands, for
debugging. Update and extend comments. Escape literal dots in
grep regular expressions.
Ralf Wildenhues [Fri, 21 Jan 2011 18:54:35 +0000 (19:54 +0100)]
tests: avoid false failure in cygnus-dependency-tracking.test.
* tests/cygnus-dependency-tracking.test: Be less restrictive
when grepping the compiler error message, GCC 3.4.6 on FreeBSD
does not mention an undefined symbol.
Ralf Wildenhues [Wed, 19 Jan 2011 20:50:02 +0000 (21:50 +0100)]
install-sh: avoid Tru64 sh `test' operator precedence issues.
* lib/install-sh: Protect file names and directory components
that consist of `=', `(', `)', or `!'. Move protection as early
as possible, to avoid errors such as with Tru64 sh `test -z ='.
* tests/instsh2.test: Extend test to cover more possibilities.
Fixes 1.12 instspc-equal-install.test failure on Tru64/OSF 5.1.
Ralf Wildenhues [Tue, 18 Jan 2011 21:58:18 +0000 (22:58 +0100)]
tests: work around Tru64 sh -e issues for instspc*.test.
Apparently, Tru64 sh does not like turning off errexit mode,
and gets confused.
* tests/defs: Document 'errexit' cleanliness requirement.
* tests/defs-static.in: Likewise. Avoid error from command
substitution to abort instspc*.test with Tru64/OSF 5.1 sh.
* tests/instspc-tests.sh: Drop now-unneeded temporary errexit
dropping. Add strategic '|| Exit' to let tests work on Tru64.
Ralf Wildenhues [Wed, 19 Jan 2011 18:46:27 +0000 (19:46 +0100)]
Allow _AM_DEPENDENCIES to be used later in configure.
* m4/depend.m4 (_AM_DEPENDENCIES): Remove a previously existing
conftest.dir before recreating it.
Fixes bug#7864.
Report by Eric Blake, from report by Scott McCreary against M4.
Ralf Wildenhues [Sun, 16 Jan 2011 21:27:37 +0000 (22:27 +0100)]
tests: allow double-colon spy.test to work with HP-UX make.
* tests/spy.test: Fix comment typos. Ensure prerequisites we
do not want to depend on are strictly older than the target.
Also test with a target out of date wrt. more than one rule.
projects / thirdparty / automake.git / log
