POSIX says that suffix rules cannot have prerequisites, but after
making this change we observed a number of makefiles "in the wild"
that were relying on this behavior and failed.
For .POSIX: makefiles, obey POSIX. Otherwise preserve the old
behavior. However, generate a warning so users know this is a
problem. In a future version we will change all behavior to be
POSIX-conforming.
* NEWS: describe the change
* src/rule.c (convert_to_pattern): If posix_pedantic don't make a
pattern rule if prereqs exist. Otherwise show a warning.
* tests/scripts/features/suffixrules: Add tests for the new behavior
including .POSIX vs. non-.POSIX.
Initial implementation by Christof Warlich <cwarlich@gmx.de>
* NEWS: Announce the new feature.
* doc/make.texi (Other Special Variables): Document .EXTRA_PREREQS.
* src/dep.h (struct dep): New flag to note extra prereq deps.
* src/filedef.h (expand_extra_prereqs): Declare a function to expand
the value of .EXTRA_PREREQS.
* src/file.c (expand_extra_prereqs): Given a struct variable lookup
of .EXTRA_PREREQS, convert it into a list of deps and for each one
make sure it has a struct file and has the new flag set.
(snap_file): A new function invoked by hash_map that will perform
per-file operations: set up second expansion, intermediate, and also
.EXTRA_PREREQS. Manage circular dependencies by ignoring them.
(snap_deps): Defer per-file operations until the end. Look up the
global .EXTRA_PREREQS and pass it along to snap_file for each file.
* src/implicit.c (struct patdeps): Remember the extra prereqs flag.
(pattern_search): Transfer extra prereqs flag settings into the
matched pattern rule.
* src/rule.h (snap_implicit_rules): Rename count_implicit_rules to
snap_implicit_rules since we now do more than count.
* src/rule.c (snap_implicit_rules): As we walk through all the pattern
rules, add in any global .EXTRA_PREREQS to the dep list. Ensure we
take them into account for the max number of prereqs and name length.
* src/main.c (main): Add extra-prereqs to .FEATURES.
Call the renamed snap_implicit_rules.
* tests/scripts/variables/EXTRA_PREREQS: Add tests.
Avoid using posix_spawn implementations that fail asynchronously when
the spawned program can't be invoked: this means instead of getting
an error such as "No such file or directory" we get just "Exit 127".
Original implementation of the configure.ac macro provided by
Martin Dorey <martin.dorey@hds.com>
Original implementation of the regression tests provided by
Dmitry Goncharov <dgoncharov@users.sf.net>
* configure.ac: Test whether posix_spawn fails asynchronously. In a
cross-compilation environment, assume that it does not. If we detect
that it does, fall back to fork/exec.
* tests/scripts/features/exec: Add regression tests for different
shebang invocation methods.
* src/dep.h: Add a new flag PARSEFS_ONEWORD
* src/read.c (parse_file_seq): If PARSEFS_ONEWORD is given, treat the
entire incoming string as a single pattern.
* src/implicit.c (pattern_search): Pass PARSEFS_ONEWORD when parsing
patterns for wildcards.
* tests/scripts/features/patternrules: Add a new test.
If the stem matches a path containing a directory not just a
filename, make sure the second expansion of $* in the
prerequisites matches $* in the recipe. This requires using
$(*F) when replacing % in the first expansion to preserve the
simple filename.
* src/implicit.c (pattern_search): If lastslash is set prepend
the directory onto the stem. Then use $(*F) when expanding %.
* tests/scripts/features/se_implicit: Add a test case
On Windows the path to the helper tool will contain '\': this will
fail if recipes are run with a POSIX shell. Convert '\' to '/'
on Windows. While here, escape any spaces in the path as well.
* tests/thelp.pl: Rename from tests/jhelp.pl.
(op): Use names instead of options for the operations.
(op): Add new operations for sleep, mkdir, and rm.
(op): Enhance wait to time out
* tests/run_make_tests.pl: Add a new #HELPER# replacement
(subst_make_string): Use fully-qualified path to thelp.pl
* tests/scripts/features/parallelism: Update to use thelp.pl
and the new named operations. Use thelp.pl sleep instead of
system-specific sleep commands.
* tests/scripts/features/output-sync: Update to use thelp.pl
instead of complex shell scripts.
* Makefile.am: Distribute tests/thelp.pl instead of tests/jhelp.pl
* tests/config-flags.pm.W32: Create a predefined Windows file.
* Makefile.am (test_FILES): Add it to the distribution.
* build_w32.bat: Install tests/config-flags.pm if not existing.
* tests/run_make_tests.pl (get_config): Create new function.
* tests/scripts/features/archives: Call get_config() rather than
using %CONFIG_FLAGS directly.
* tests/scripts/features/load: Ditto.
* tests/scripts/features/loadapi: Ditto.
* tests/scripts/functions/wildcard: Ditto.
This reverts commit 6264deece3.
Further investigation discovers that the real issue is that
GNU Emacs compile mode doesn't have a matching regex for GNU
make error messages generated when targets fail. I submitted
a patch to GNU Emacs adding a matcher for compile mode.
Ensure we properly reduce job_slots_used if a command fails because
it doesn't exist/can't be started.
* src/job.h (struct child): Add a field jobslot to be set when using
a job slot.
* src/job.c (start_waiting_job): Remember if we are using a job slot.
(reap_children): Reduce number of job slots used by jobslot.
Go through both run_make_tests.pl and test_driver.pl and slightly
modernize the Perl and clean up indentation etc. Fix a number of
warnings in the test scripts detected by running with -w.
* tests/test_driver.pl: Move make error string detection out of the
base test driver.
(run_all_tests): Ensure that we always look for tests in the cwd.
* tests/run_make_tests.pl: Use File::Spec for path manipulations.
Correctly use setlocale() when detecting error strings.
Get configuration from the config-flags.pm file not config.status.
* tests/scripts/features/archives: Use new $cwddir variable.
* tests/scripts/features/reinvoke: Add missing semicolon.
* tests/scripts/features/vpath2: Avoid non-existent variable.
* tests/scripts/functions/foreach: Escape variables.
* tests/scripts/misc/bs-nl: Remove non-existing \v escape sequence.
* tests/scripts/misc/general4: Use handy create_file().
* tests/scripts/options/dash-C: Use Cwd/$cwddir.
* tests/scripts/options/dash-I: Use subst_make_string() and #PWD#.
* tests/scripts/options/symlinks: Use File::Spec.
* tests/scripts/targets/DEFAULT: Use create_file and run_make_test.
* tests/scripts/variables/CURDIR: Use run_make_test.
* tests/scripts/variables/automatic: Remove extraneous "\".
* tests/scripts/vms/library: Remove extra "my" and extraneous "\".
When using exec we install the child's environment before invoking
execlp(), so commands are found on the child's PATH. posix_spawnp
searches on the parent's PATH, which we don't want.
Import gnulib's findprog-in module and use it to search the child's
PATH, then use posix_spawn() to run it.
Also, posix_spawn() does not fall back to trying sh on ENOEXEC, as
execlp() does, so implement that as well.
* bootstrap.conf: Add the findprog-in gnulib module
* src/job.c: Include findprog.h if we're using posix_spawn.
(start_job_command): Remove the handling of child->cmd_name,
(child_execute_job): and add it here. Look up the command to be
run in the child's path and invoke it if found. If it fails with
ENOEXEC then retry it as an argument to the default shell.
* tests/scripts/misc/general4: Test makefile PATH assignments.
* tests/scripts/features/targetvars: Ditto, for target variables.
The built-in glob implementation does not correctly handle dangling
symlinks. This needs to be fixed by switching to the latest glob
implementation from gnulib but that's a big job: for now avoid the
test if we know it will fail.
The check for matching a file's command to the default rule's command
does not account for null. If no .DEFAULT is defined a rule with no
recipe has it's $< variable set to the value of $@. This breaks second
expansion, particularly when used with pattern rules.
* src/commands.c [set_file_variables]: Check that cmds is set
* tests/scripts/features/se_explicit: Test case
* tests/scripts/features/se_implicit: Test case
Copyright-paperwork-exempt: yes
Ensure that variable overrides are passed to recursive make instances
even if no overrides were provided on the command line.
Fix suggested by Rici Lake <ricilake@gmail.com>
* src/main.c (define_makeflags): Add overrides without respect to the
value of command_variables.
* tests/scripts/features/recursion: Add a test.
When merging the variable_set_lists for two targets it could be
that the "from" set list is a subset of the "to" set list: check
for this situation to avoid freeing used memory.
* src/variable.c (merge_variable_set_lists): Walk the "to" list and
if the "from" list is contained in it, nothing to do.
* tests/scripts/features/se_explicit: Add a test.
Create a new variable run_silent to hold the current instance's global
silence setting, allowing silent_flag to represent only whether the -s
option was provided on the command line.
* src/makeint.h: Change silent_flag variable to run_silent.
* src/job.c: Ditto.
* src/remake.c: Ditto.
* src/file.c: Ditto.
* src/main.c: Add a new global variable run_silent.
(decode_switches): After switches are decoded, initialize run_silent.
* tests/scripts/targets/SILENT: Add a test for recursive behavior.
This patch allows "grouped targets" using the &: syntax:
tgt1 tgt2 ... tgtn &: pre1 pre2 ...
recipe
When the &: separator is used (in single or double colon forms), all
the targets are understood to be built by a single invocation of the
recipe. This is accomplished by piggy-backing on the already-existing
pattern rule feature, using the file's "also_make" list.
* NEWS: Add information about grouped targets.
* doc/make.texi (Multiple Targets): Add information on grouped targets.
(Pattern Intro): Refer to the new section to discuss multiple patterns.
* src/main.c (main): Add "grouped-targets" to .FEATURES
* src/read.c (make_word_type): Add new types for &: and &::.
(eval): Recognize the &: and &:: separator and remember when used.
(record_files): Accept an indicator of whether the rule is grouped.
If so, update also_make for each file to depend on the other files.
(get_next_mword): Recognize the &: and &:: word types.
* tests/scripts/features/grouped_targets: New test script.
* AUTHORS: Add Kaz Kylheku
Fixes an issue seen in the Linux kernel build system, reported by
Masahiro Yamada <yamada.masahiro@socionext.com>. Fix suggested
on the bug-make mailing list by Mike Shal <marfey@gmail.com>.
* tests/scripts/features/parallelism: Add a test to verify this.
* NEWS: Update with a backward-compatibility warning.
* src/rule.c (convert_to_pattern): If a suffix rule has dependencies,
do not create a pattern rule for it. According to the manual suffix
rules with prerequisites are treated as normal targets.
* tests/scrips/features/suffixrules: Create some regression tests for
.SUFFIXES and suffix rules.
If multiple pattern rules have the same pattern as also-make targets
and we attempt to run them at the same time, we might downgrade the
command state from 'running' to 'deps_running'; this will prevent
that also_make from being considered complete causing make to wait
forever for it to finish.
Ensure that set_command_state never downgrades the state of a target.
* src/file.c (set_command_state): Don't downgrade command_state.
* src/filedef.h (struct file): Document the order prerequisite.
* test/scripts/features/patternrules: Test the behavior.
* src/function.c (func_shell_base): Use error() instead of recreating
the error output.
* src/job.c (exec_command): Show more standard error messages.
* src/load.c (unload_file): Fix whitespace in the error message.
* tests/scripts/features/errors: Add tests for starting non-
existent commands and new error message formats.
* tests/scripts/features/output-sync: New error message formats.
* tests/scripts/functions/shell: Ditto.
If we failed to fork() we were essentially exiting make immediately
without respect to ignore flags, etc. On one hand that makes sense
because if you can't fork you're in real trouble, but it doesn't
work so well on systems where we don't fork at all. Instead, treat
a fork error like any other error by delaying the handling until
the next call to reap_children(). Any child with a PID of -1 is
considered to have died before starting so check these first without
waiting for them.
* src/commands.c (fatal_error_signal): Don't kill children that
never started.
* src/function.c (func_shell_base): Handle cleanup properly if the
child doesn't start.
* src/job.c (reap_children): Check for children that died before
starting and handle them without waiting for the PID.
(start_job_command): Free memory when the child doesn't start.
(start_waiting_job): Don't manage children who never started.
(child_execute_job): If the fork fails return PID -1.
* src/vmsjobs.c: Check for children that never started.
* tests/run_make_tests.pl: Parse config.status to get all options.
* tests/scripts/features/jobserver: Windows doesn't use pipes
* tests/scripts/functions/shell: Don't test kill -2 on Windows
* tests/scripts/misc/bs-nl: Windows doesn't handle single quotes
* tests/scripts/misc/general3: Ditto.
Move the source code (other than glob) into the "src" subdirectory.
Update all scripting and recommendations to support this change.
* *.c, *.h, w32/*: Move to src/
* configure.ac, Makefile.am, maintMakefile: Locate new source files.
* Basic.mk.template, mk/*: Update for new source file locations.
* NEWS, README.DOS.template: Update for new locations.
* build.template, build_w32.bat, builddos.bat: Ditto.
* po/POTFILES.in: Ditto
* tests/run_make_tests.pl, tests/scripts/features/load*: Ditto.
* make.1: Move to doc.
* mk/VMS.mk: Add support for building on VMS (hopefully).
* makefile.vms, prepare_w32.bat: Remove.
* SCOPTIONS: Update to define HAVE_CONFIG_H
Over time the non-standard build and install systems (nmake files,
smake files, Visual Studio project files, etc.) have atrophied and
maintaining them is not worth the effort, for such a simple utility
as make. Remove all the non-standard build tool support and unify
OS-specific build rules under a basic set of (GNU make) makefiles.
Preserve the existing bootstrapping scripts (for POSIX, Windows,
and MS-DOS). Also the existing VMS build scripts are left unchanged:
I don't have enough experience with VMS to venture into this area.
Perhaps one of the VMS maintainers might like to determine whether
conversion would be appropriate.
Rather than create libraries for w32 and glob (non-POSIX), simply
link the object files directly to remove the complexity.
* NEWS: Update with user-facing notes.
* Makefile.am: Clean up to use the latest automake best practices.
Build Windows code directly from the root makefile to avoid recursion.
* README.Amiga, README.DOS.template, README.W32.template: Updated.
* INSTALL: Point readers at the README.git file.
* maintMakefile: Remove obsolete files. Create Basic.mk file.
* Basic.mk.template, mk/*.mk: Create basic GNU make-based makefiles.
* build_w32.bat: Copy Basic.mk to Makefile
* configure.ac: We no longer need AM_PROG_AR.
* dosbuild.bat: Rename to builddos.bat. Incorporate configure.bat.
* Makefile.DOS.template: Remove.
* NMakefile.template, w32/subproc/NMakefile: Remove.
* SMakefile.template, glob/SMakefile, glob/SCOPTIONS, make.lnk: Remove.
* configure.bat, glob/configure.bat: Remove.
* w32/Makefile.am: Remove.
* make_msvc_net2003.sln, make_msvc_net2003.vcproj: Remove.
This is about twice as fast as the current hash, and removes the
need for double hashing (improving locality of reference). The
hash function is based on Bob Jenkins' design, slightly adapted
wherever Make needs to hash NUL-terminated strings. The old hash
function is kept for case-insensitive hashing.
This saves 8.5% on QEMU's no-op build (from 12.87s to 11.78s).
* configure.ac: Check endianness.
* hash.c (rol32, jhash_mix, jhash_final, JHASH_INITVAL,
sum_get_unaligned_32, jhash): New.
* hash.h (STRING_HASH_1, STRING_N_HASH_1): Use jhash.
(STRING_HASH_2, STRING_N_HASH_2): Return a dummy value.
(STRING_N_COMPARE, return_STRING_N_COMPARE): Prefer memcmp to strncmp.
* variable.h (enum variable_flavor: Add a new flavor for appended
values that shouldn't be expanded.
* variable.c (do_variable_definition): If given this new flavor,
do not expand the value before appending it.
* read.c (eval_makefile): Use this new flavor for MAKEFILE_LIST
* tests/scripts/variables/MFILE_LIST: Test filenames containing '$'.
* doc/make.texi (Appending): Document this behavior.
* variable.c (do_variable_definition): Only add a space if the variable
value is not empty.
* tests/scripts/variables/flavors: Test this behavior.
* main.c (main): Sanitize program name detection on Windows.
* makeint.h: 'program' is a const string on all platforms now.
* tests/run_make_tests.bat: Windows bat file to invoke tests
* tests/test_driver.pl: Obtain system-specific error messages.
(get_osname): Compute the $port_type here. Add more $osname checks
for different Windows Perl ports.
(_run_command): Rewrite the timeout capability to work properly
with Windows. Don't use Perl fork/exec; instead use system(1,...)
which allows a more reliable/proper kill operation.
Also, allow options to be given as a list instead of a string, to
allow more complex quoting of command-line arguments.
* tests/run_make_tests.pl (run_make_with_options): Allow options
to be provided as a list in addition to a simple string.
(set_more_defaults): Write sample makefiles and run make on them
instead of trying to run echo and invoking make with -f-, to avoid
relying on shell and echo to get basic configuration values. Also
create a $sh_name variable instead of hard-coding /bin/sh.
* tests/scripts/features/archives: Skip on Windows.
* tests/scripts/features/escape: Use list method for passing options.
* tests/scripts/features/include: Use system-specific error messages.
* tests/scripts/features/output-sync: "Command not found" errors
generate very different / odd output on Windows. This needs to be
addressed but for now disable these tests on Windows.
* tests/scripts/functions/abspath: Disable on Windows.
* tests/scripts/functions/file: Use system-specific error messages.
* tests/scripts/functions/shell: "Command not found" errors generate
very different / odd output on Windows. This needs to be addressed
but for now disable these tests on Windows.
* tests/scripts/misc/close_stdout: Disable on Windows.
* tests/scripts/options/dash-k: Use system-specific error messages.
* tests/scripts/options/dash-l: Disable on Windows.
* tests/scripts/options/eval: Use list method for passing options.
* tests/scripts/options/general: Skip some non-portable tests.
* tests/scripts/targets/ONESHELL: Skip some non-portable tests.
* tests/scripts/targets/POSIX: Skip some non-portable tests.
* tests/scripts/variables/MAKEFILES: Skip some non-portable tests.
* tests/scripts/variables/SHELL: Use a makefile not -f- for testing.
* make.1: Document the -E and --eval options.
* doc/make.texi: Document the -E option.
* tests/scripts/options/eval: Test the -E option and MAKEFILES.
* NEWS: Add information about the new option.
* make.1: Document the new flag.
* doc/make.texi: Document the new flag. Remove suggestions that the
.SILENT special target is deprecated or should not be used.
* tests/scripts/options/dash-s: Test the -s and --no-silent options.
* NEWS: Add information about the new option.
* read.c (eval_makefile): Set deps->error if we discovered any
error reading makefiles, and set NONEXISTENT_MTIME so we know
it needs to be rebuilt.
* main.c (main): Clean up management of makefile_mtimes.
* tests/scripts/features/include: Add open failure testcases.
* tests/test_driver.pl: Save error strings for later comparison.
* tests/run_make_tests.pl: Create portable commands for later use.
* tests/*: Use these new variables.
* NEWS: Document the change, as a backward-incompatible change.
* main.c (main): Add 'nocomment' to the .FEATURES variable.
* read.c (remove_comments): Skip variable references during remove.
(find_char_unquote): Fix comments for new STOPMAP support.
* tests/scripts/features/escape: Test new escape syntax.
* tests/scripts/functions/guile: Ditto.
* tests/scripts/functions/shell: Ditto.
The fix for SV 44742 had a side-effect that some double-colon targets
were skipped. This happens because the "considered" facility assumed
that all targets would be visited on each walk through the dependency
graph: we used a bit for considered and toggled it on each pass; if
we didn't walk the entire graph on every pass the bit would get out
of sync. The new behavior after SV 44742 might return early without
walking the entire graph. To fix this I changed the considered value
to an integer which is monotonically increasing: it is then never
possible to incorrectly determine that a previous pass through the
graph already considered the current target.
* filedef.h (struct file): make CONSIDERED an unsigned int.
* main.c (main): No longer need to reset CONSIDERED.
* remake.c (update_goal_chain): increment CONSIDERED rather than
inverting it between 0<->1.
(update_file_1): Reset CONSIDERED to 0 so it's re-considered.
(check_dep): Ditto.
* tests/scripts/features/double_colon: Add a regression test.
* remake.c (update_file): Don't update double-colon target status
if we're still building targets.
(ftime_t): Don't propagate timestamps for double-colon targets that
we've not examined yet.
* tests/scripts/features/double_colon: Add parallel build tests.
Copyright-paperwork-exempt: yes
While displaying line numbers, show the relevant line number inside
the recipe not just the first line of the entire recipe.
Sample changes suggested by Brian Vandenberg <phantall@gmail.com>
* gnumake.h (gmk_floc): Add an 'offset' to track the recipe offset.
* read.c (eval, eval_makefile, eval_buffer): Initialize 'offset'.
(record_files, install_pattern_rule): Ditto.
* job.c (new_job, job_next_command): Update 'offset' based on the
line of the recipe we're expanding or invoking.
(child_error): Add 'offset' when showing the line number.
* function.c (func_shell_base): Ditto.
* output.c (error, fatal): Ditto.
* NEWS: Mention the new ability.
* tests/scripts/features/errors: Check the line number on errors.
* tests/scripts/functions/warning: Check the line number on warnings.
* tests/scripts/features/output-sync,
tests/scripts/features/parallelism, tests/scripts/functions/shell,
tests/scripts/functions/error: Update line numbers.
Delay the generation of error messages for included files until we
are sure that we can't rebuild that included file.
* dep.h (struct dep): Don't reuse "changed"; make a separate field
to keep "flags". Get rid of dontcare and use the flag.
(struct goaldep): Create a new structure for goal prereqs
that tracks an errno value and the floc where the include happened.
Rework the structures to ensure they are supersets as expected.
In maintainer mode with GCC, use inline to get type checking.
* read.c (eval_makefile): Return a struct goaldep for the new
makefile. Ensure errno is set properly to denote a failure.
(read_all_makefiles): Switch to goaldep and check errno.
(eval): Don't show included file errors; instead remember them.
* remake.c (update_goal_chain): Set global variables to the current
goaldep we're building, and the entire chain.
(show_goal_error): Check if the current failure is a consequence
of building an included makefile and if so print an error.
(complain): Call show_goal_error() on rule failure.
* job.c (child_error): Call show_goal_error() on child error.
* main.c (main): Switch from struct dep to goaldep.
* misc.c (free_dep_chain): Not used; make into a macro.
* tests/scripts/features/include: Update and include new tests.
* tests/scripts/options/dash-B, tests/scripts/options/dash-W,
tests/scripts/options/print-directory,
tests/scripts/variables/MAKE_RESTARTS: Update known-good-output.
Previously if the jobserver was active, MAKEFLAGS would contain only
the -j option but not the number (not -j5 or whatever) so users
could not discover that value. Allow that value to be provided in
MAKEFLAGS without error but still give warnings if -jN is provided
on the command line if the jobserver is already activated.
* NEWS: Discuss the new behavior.
* os.h, posixos.c, w32/w32os.c: Return success/failure from
jobserver_setup() and jobserver_parse_auth().
* main.c (main): Separate the command line storage of job slots (now
in arg_job_slots) from the control storage (in job_slots). Make a
distinction between -jN flags read from MAKEFLAGS and those seen
on the command line: for the latter if the jobserver is enabled then
warn and disable it, as before.
* tests/scripts/features/jobserver: Add new testing.
* NEWS: Mention the change.
* main.c: Rename jobserver_fds variable to jobserver_auth and
--jobserver-fds option to --jobserver-auth.
* os.h, posixos.c, w32/w32os.c: Rename jobserver_parse_arg() and
jobserver_get_arg() to jobserver_parse_auth()/jobserver_get_auth().
* makeint.h: Change MAP_SPACE to MAP_NEWLINE, and add MAP_PATHSEP
and MAP_SPACE which is now MAP_BLANK|MAP_NEWLINE. Create
NEW_TOKEN(), END_OF_TOKEN(), ISBLANK(), ISSPACE() macros.
* main.c (initialize_stopchar_map): Set MAP_NEWLINE only for
newline characters.
* Convert all uses of isblank() and isspace() to macros.
* Examine all uses of isblank() (doesn't accept newlines) and
change them wherever possible to ISSPACE() (does accept newlines).
* function.c (func_foreach): Strip leading/trailing space.
* variable.c (parse_variable_definition): Clean up.
* tests/scripts/functions/foreach: Test settings and errors.
* tests/scripts/functions/call: Rewrite to new-style.
* tests/scripts/misc/bs-nl: Add many more tests for newlines.
* NEWS: Add information about reading files.
* make.texi (File Function): Describe reading files.
* tests/scripts/functions/file: Test new features for $(file ...)
For performance, we only recompute .VARIABLES when (a) it's expanded
and (b) when its value will change from a previous expansion. To
determine (b) we were checking the number of entries in the hash
table which used to work until we started undefining entries: now if
you undefine and redefine the same number of entries in between
expanding .VARIABLES, it doesn't detect any change. Instead, keep
an increasing change number.
* variables.c: Add variable_changenum.
(define_variable_in_set, merge_variable_sets): Increment
variable_changenum if adding a new variable to the global set.
(undefine_variable_in_set): Increment variable_changenum if
undefining a variable from the global set.
(lookup_special_var): Test variable_changenum not the hash table.
* tests/scripts/variables/special: Test undefining variables.
* main.c (main): Pre-define .LOADED as a default-level variable.
* load.c (load_file): Set the value rather than append it. Avoid
adding an extra initial whitespace.
* tests/scripts/features/load: Run with --warn-undefined-variables.
* job.h (struct child): New bit to mark recursive command lines.
* job.c (start_job_command): Set the recursive command line bit.
(reap_children): If the child is a recursive command and it exits
with 1 during question mode, don't print an error and exit with 1.
* tests/scripts/options/dash-q: Add a regression test.
Newer versions of binutils allow ar to be compiled to generate
"deterministic archives" by default: in this mode no timestamp
information is generated in the static archive, which utterly
breaks GNU make's archive updating capability. Debian and Ubuntu
have turned this feature on by default in their distributions
which causes the regression tests to fail.
Update the regression tests to check for the availability of the
"U" option to ar which disables deterministic archives and allows
GNU make's archive support to work properly again.
In order to fix SV 12267 we were marking the prerequisites of
implicit (pattern) targets that existed elsewhere in the makefile
as precious to keep them from being deleted as intermediate files.
However this also keeps them from being deleted on error. Instead
mark them as secondary.
* tests/scripts/targets/DELETE_ON_ERROR: Test DELETE_ON_ERROR.
