The POSIX standard only requires false(1) to return a "non-zero" exit
code; almost all systems return 1 but some (Solaris!!!) return 255 or
possibly even other values. Use our helper "fail" instead.
* tests/thelp.pl: Have the "fail" command obey -q.
* tests/scripts/features/parallelism: Helper -q no longer prints fail.
* tests/scripts/targets/POSIX: Replace false with #HELPER# -q fail 1.
* tests/scripts/variables/MAKEFLAGS: Ditto.
* tests/scripts/variables/SHELL: Ditto.
Using backslashes in a sed command line is tricky as different programs
use them differently as escape sequences. Eli Zaretskii points out
that Windows "echo" doesn't do any processing, so rework all our sed
invocations to use script files created by echo.
* bootstrap.bat: Use echo to create sed script files instead of -e.
* Basic.mk.template: Fix typo in the comments.
* .gitignore: Ignore any .sed scripts.
Don't generate undefined variable warnings for variables that are
internal / special to make and where the empty string is valid.
Rather than defining them to empty, which could introduce unwanted
behavior, keep a list of variable names which we should never warn
about.
* src/variable.h (warn_undefined): Convert the macro to a function.
* src/variable.c (defined_vars): Always "defined" variable names.
(warn_undefined): Implement as a function and check against the
defined_vars before generating a warning.
* src/read.c (read_all_makefiles): No need to reset warning flag.
* src/vpath.c (build_vpath_lists): Ditto.
* tests/scripts/options/warn-undefined-variables: Expand all the
pre-defined variables to ensure warnings are not generated.
* src/makeint.h (reset_makeflags): New function to handle changing
MAKEFLAGS from within makefiles. Remove decode_env_switches().
* src/variable.c (set_special_var): Call reset_makeflags() instead
of various internal methods.
* src/main.c (decode_env_switches): Only internal now so make static.
(decode_switches): Don't invoke construct_include_path() yet.
(reset_makeflags): Decode env switches and construct include paths.
(main): Construct include paths after we process -C options.
* tests/scripts/options/dash-C: Rewrite to use new test constructs.
Add a test using both -C and -I together.
Add a test for multiple -C options.
Strawberry Perl has some different behaviors from ActiveState Perl
which impact the test suite:
- Avoid Perl's chomp() as it may not remove CRs; chomp() may remove
only the final NL but not the CR in a CRNL line ending.
- Strawberry Perl doesn't support ActiveState's system(1, ...) form.
- Strawberry Perl (or msys?) does something weird with "/tmp" when
provided to exec(), replacing it with the user's %TEMP%.
- Strawberry Perl uses msys paths like /c/foo instead of C:\foo.
* tests/test_driver.pl (get_osname): Strawberry Perl uses 'msys' as
its $^O so if we see that use a port of 'W32'.
(_run_with_timeout): Strawberry Perl doesn't support the special
system(1, ...) form of system() so use POSIX standard fork/exec.
(compare_answer): Paths generated by Strawberry Perl use msys path
format (e.g., /c/foo instead of C:\foo); check for those differences
and compare RE against both the unmodified and modified log.
* tests/run_make_tests.pl (set_defaults): Switch from chomp to s///
to remove CRNL and NL line endings.
* tests/scripts/features/errors: Executing directories on Strawberry
will give an error; translate it to Windows error output format.
* tests/scripts/features/output-sync: Ditto.
* tests/scripts/features/temp_stdin: Ditto.
* tests/scripts/functions/realpath: Ditto.
* tests/scripts/options/dash-I: Ditto.
* tests/scripts/variables/INCLUDE_DIRS: Ditto.
* tests/scripts/misc/close_stdout: /dev/full is reported as existing
on Strawberry Perl, but it doesn't do anything. Skip the test.
* tests/scripts/variables/MAKEFLAGS: When an argument containing
/tmp is passed to a program via exec(), something replaces it with
the expansion of the %TEMP% variable. Instead of using /tmp create
a local directory to use.
Sometimes the error message is expected to contain a newline, other
times it is not. Also the result of FormatMessage() sometimes ends
in CRNL already: when printed via fprintf() the final newline is
converted to CRNL, giving an output of CRCRNL which causes tests
to fail to match. Remove any CR/NL chars from the result string.
* src/job.c (reap_children): Add \n to the error message fprintf.
(exec_command): Ditto.
* src/w32/subproc/w32err.c (map_windows32_error_to_string): Remove
any trailing CR or NL from the string before returning.
* doc/make.texi (Options/Recursion): Clarify that MAKEFLAGS values
from the environment have precedence over those set in the makefile.
* tests/scripts/variables/MAKEFLAGS: Check boolean switches -k/-S,
-w/--no-print-directory and -s/--no-silent as follows:
1. A switch can be enabled or disabled on the command line.
2. A switch can be enabled or disabled in env.
3. A switch can be enabled or disabled in makefile.
4. Command line beats env and makefile.
5. Env beats makefile.
6. MAKEFLAGS contains each specified switch at parse and build time.
7. If switches are specified in multiple origins, MAKEFLAGS contains
the winning switch at parse and build time.
8. MAKEFLAGS does not contain the losing switch.
Also test that --debug settings from different origins are combined
together into one option.
Certain switches, such as -S or --no-silent, turn on behavior that is
enabled by default. When a switch is specified via the command line,
makefile, or env, ensure the switch is added to MAKEFLAGS.
* src/main.c (struct command_switch): Add bit "specified".
(switches): Initialize command_switch->specified.
(decode_switches): Set command_switch->specified.
(define_makeflags): Check command_switch->specified.
Certain options can be flipped on and off: -k/-S, -s/--no-silent, and
-w/--no-print-directory. Ensure they behave as follows:
1. A switch can be enabled or disabled on the command line.
2. A switch can be enabled or disabled in env.
3. A switch can be enabled or disabled in makefile.
4. Command line beats env and makefile.
5. Env beats makefile.
* src/main.c: Add variables to hold the origin of relevant options.
(struct command_switch): Add origin field.
(switches): Set a pointer to hold the origin of relevant options.
(decode_switches): For any switch that can be specified in makefile or
env, honor the switch only if cs->origin is not set or the specified
origin beats cs->origin. Set cs->origin when relevant.
* src/makeint.h: Replace print_directory flag with should_print_dir().
* src/main.c (main): Remove print_directory flag and related code.
(should_print_dir): Create.
* src/output.c (output_dump): Use should_print_dir().
(output_start): Ditto.
Ensure included makefiles are not treated as intermediate, even if
they are created by an implicit rule.
Reported by Patrick Oppenlander <patrick.oppenlander@gmail.com>.
* src/read.c (eval_makefile): Mark makefiles as explicit.
* tests/scripts/features/include: Add a test.
* src/makeint.h (HAVE_DRIVESPEC): Create a macro to check.
* src/main.c (.FEATURES): Add "dospaths" as a feature.
* src/read.c (eval_makefile) [DOS]: If the included makefile name
starts with a drivespec, don't search the include directories.
* doc/make.texi (Include): Document this behavior.
* tests/scripts/features/include: Add a test.
Fix .NOTINTERMEDIATE without prerequisites to disable intermediate
status for all targets.
* src/makeint.h: Declare extern no_intermediates.
* src/main.c: Add global definition of no_intermediates.
* src/file.c: Remove static no_intermediates to use global variable.
(remove_intermediates): Check no_intermediates.
* src/implicit.c (pattern_search): For a file found by implicit search
set file->notintermediate if no_intermediates is set.
* src/remake.c (update_file_1): Don't set file->secondary for a
pre-existing file if no_intermediates is set. The check for
no_intermediates here is redundant, but won't hurt: keep it in case
things change so that it matters.
* tests/scripts/targets/NOTINTERMEDIATE: Fix a test.
This commit introduces two visible changes:
1. Keep command line variable assignments in MAKEFLAGS at all times,
even while parsing makefiles.
2. Define makeflags immediately when a makefile modifies MAKEFLAGS.
The new MAKEFLAGS and MAKEOVERRIDES initialization procedure:
1. decode_switches (argc, argv, o_command) is called to parse command
line variable assignments.
2. Command line variable assignments go through quote_for_env.
Initialize -*-command-variables-*- to the quoted values.
3. MAKEOVERRIDES is initialized to refer to -*-command-variables-*-
with origin o_env to keep the definitions in the database intact.
4. define_makeflags() is called which adds MAKEOVERRIDES to MAKEFLAGS.
5. Makefiles are parsed. If a makefile modifies MAKEFLAGS, the new
value of MAKEFLAGS is defined right away.
6. Env switches are decoded again as o_env. The definitions set by
decode_switches at step 1 stay intact, as o_command beats o_env.
We must preserve the original intact definitions in order to detect
failure cases; for example:
$ cat makefile
all:; $(hello)
$ make hello='$(world'
makefile:1: *** unterminated variable reference. Stop.
* src/makeint.h: Declare enum variable_origin, struct variable and
define_makeflags(). Add parameter origin to decode_env_switches().
* src/main.c (define_makeflags): Remove "all". If a variable is
assigned on the command line then append MAKEOVERRIDES to MAKEFLAGS.
(decode_env_switches): Replace parameter env with origin.
(decode_switches): Replace parameter env with origin.
Treat origin == o_command as env == 0.
(handle_non_switch_argument): Replace parameter env with origin.
Treat origin == o_command as env == 0.
(main): Call decode_switches() with origin==o_command before parsing
makefiles. Call decode_switches() with origin==o_env after parsing
makefiles.
* src/variable.c (set_special_var): Define makeflags at parse time,
each time a makefile modifies MAKEFLAGS.
(do_variable_definition): Strip command line variable assignments from
MAKEFLAGS before appending extra flags. set_special_var() adds them
back.
* tests/scripts/variables/MAKEFLAGS: Add tests.
* src/read.c (eval): If "missing separator" appears to be due to
missing space after ifeq/ifneq, give a hint about the error.
* tests/scripts/misc/failure: Check for these types of failures.
* tests/scripts/variables/special: Move error checking unrelated
to special variables, to misc/failure.
If make cannot create a temporary lock file for output sync, continue
without output sync enabled rather than dying.
However, if make cannot store a makefile from stdin to a temporary
file that is still a fatal error.
* misc.c (get_tmppath): Keep running on failure to generate a
temporary file name.
(get_tmpfd): Keep running on failure to get a temporary file.
(get_tmpfile): Keep running on failure to open a temporary file.
Ensure memory is freed if we return an error.
* posixos.c (os_anontmp): Keep running on failure to open an
anonymous temporary file.
* output.c (setup_tmpfile): Print an error on failure to create an
output sync lock file.
* main.c (main): Die on failure to store makefile from stdin to a
temporary file.
* tests/scripts/features/output-sync: Add tests.
* tests/scripts/features/temp_stdin: Ditto.
* src/posixos.c (jobserver_parse_auth): Don't invoke fatal() if we
can't connect to an existing jobserver: just keep going without it.
* src/w32/w32os.c (jobserver_parse_auth): Ditto.
* tests/scripts/features/jobserver: Add a test for invalid FIFO
auth files.
If we fail to read a directory show the directory name in the error
message. Pass struct directory instead of directory_contents to
allow that.
(dir_file_exists_p): Change dir_contents_file_exists_p caller.
(open_dirstrem): Ditto.
Tests that try to kill the make process were not behaving as expected
on OpenBSD: the signal was sent from make to its children but the
sleep didn't die. Something odd about the way the shell treats TERM.
To reduce platform dependencies add "term" to the helper tool and run
that instead of kill / sleep.
* tests/thelp.pl: Add a new operation "term" that takes a PID.
* tests/scripts/features/output-sync: Use it.
* tests/scripts/features/temp_stdin: Ditto.
* src/function.c (define_new_function): Don't keep a pointer to the
user-provided name of a user-defined function: if the .so is unloaded
it will point to garbage. Add the name to the strcache instead.
When tests compare the output they will try converting backslashes
to slashes to see if that works. When we compare using regex's,
we can't do that because backslashes can escape special characters.
* tests/test_driver.pl (compare_output): Clean up this function.
(compare_answer_vms) [VMS]: Comparing answers on VMS is complex;
move all of it into its own function returning 0/1.
(compare_answer): A new function to compare answers: return 0/1.
Remember the CRLF->LF conversion forever; only check \ -> / when
we compare strings, not regex's.
* src/config.h.W32 (strtoll, strtoull) [__TINYC__]: Redirect to
_strtoi64 and _strtoui64, respectively. Reported by Christian
Jullien <eligis@orange.fr>.
* tests/test_driver.pl: Preserve the LSAN_OPTIONS variable.
* tests/scripts/targets/ONESHELL: Don't set a local variable.
* tests/scripts/functions/let: Test empty let variable.
* src/posixos.c (osync_parse_mutex): Free existing osync_tmpfile.
* src/misc.c (get_tmpfd): Set umask() before invoking mkstemp().
* src/ar.c (ar_parse_name): Check invalid name (shouldn't happen).
* src/function.c (define_new_function): Free previous function entry
when replacing it with a new one.
* src/job.c (child_execute_job): Initialize pid for safety.
(construct_command_argv_internal): In oneshell mode ensure that the
returned argv has the right format (0th element is a pointer to the
entire buffer).
Although gnulib requires C99, most of the code does compile with a
C90 compiler (perhaps with a lot of warnings). Reinstate our C90
configuration test, and clean up a few C99 things that crept in.
* src/job.c (construct_command_argv_internal): Don't use loop-local
variables or C++ comments.
* src/read.c (eval_makefile): Don't use loop-local variables.
This was added in Perl 5.8 but some systems still only provide older
versions such as Perl 5.6. We don't really need it anyway.
Paul Eggert <eggert@cs.ucla.edu> reported this issue.
* tests/README: Update this to be a bit more modern.
* tests/test_driver.pl: Delete the $TEMPDIR variable.
* tests/scripts/features/temp_stdin: Use $temppath not $TEMPDIR.
We attempt to do this with POSIX::setlocale() but apparently on some
systems (AIX) this isn't sufficient. So, in addition force the LC
environment variables to use "C".
Reported by Dmitry Goncharov <dgoncharov@users.sf.net>.
* tests/run_make_tests.pl: Move the global setup into set_default().
Force the %ENV locale variables to use the ones we'll use when running
make, then reset them back again after we find error messages.
Using C shells (csh, tcsh) with make is known to be problematic due
to incorrect ways it handles open file descriptors, at least. If
the user's shell is *csh then don't try it during exec tests.
It seems that some of the test environments hit the 5s timeout on
some tests. Since it doesn't really matter, as long as we don't
hang forever, increase the timeout to 60s.
* tests/test_driver.pl: Increase $test_timout to 60. We don't need
to handle VMS timeouts specially anymore.
* tests/scripts/features/parallelism: We don't need to override the
default timeout anymore.
* tests/scripts/features/patternrules: Remove confusing comment.
If we fail during setup_tmpfile() we'll try to write an error, which
will invoke setup_tmpfile() again, etc. Avoid infinite recursion.
Original patch by Dmitry Goncharov <dgoncharov@users.sf.net>
* src/output.c (setup_tmpfile): Remember we're in this function and
return immediately if we enter it during recursion.
(message): Remember the starting location and use that instead of
fmtbuf.buffer.
(error): Ditto.
(fatal): Ditto.
