1 * Document why putting @FOO@ in _SOURCES doesn't work.
2 Make it an error when the user tries to do this
3 This must be done for 1.5
5 * if automake -a, and dangling links exist, automake should
8 * add a way to require C++ support even when no apparent C++ code
9 exists. Likewise for other langs.
11 * in depend2.am, in specialization case, what if @SOURCE@ is found
12 in srcdir? We can't depend on $<! We must search explicitly.
14 * Add a way to say that foo/bar.h should be installed as
15 $(includedir)/foo/bar.h without introducing a new dir var
17 * distcheck should make sure that each file that uses _() is
20 # Verify that all source files using _() are listed in po/POTFILES.in.
22 grep -E -v '^(#|$$)' po/POTFILES.in | sort > $@-1
23 grep -E -l '\b_\(' lib/*.c src/*.c | sort > $@-2
27 * support prog_LIBS as override for LIBS
29 * serious bug. this doesn't work:
32 why did I think it did work?
34 * Sometimes we can warn about a missing `compile' script twice.
36 * A target named `INSTALL' will screw things up.
39 * Scan configure.in using the same trick that autoheader uses.
40 This will be much more reliable.
42 * Test subdir-objects option with yacc, lex, ansi2knr
43 Our locking scheme won't prevent a parallel make from losing
44 if there are two `bar.o' files and the timing is just right
45 This only happens with parallel make and no-`-c -o' compiler,
46 so it probably isn't very important
47 `-c -o' when doing libtool
48 try to find a losing compiler and see if it really works.
49 (actually: hack config.cache and do it)
51 * We're using `$<' in explicit rules when using per-exe flags
52 per-exe flags don't work for CPPFLAGS/YFLAGS/LFLAGS. Fix.
53 LIBOBJS shouldn't be used when there are per-exe flags (?)
55 * Need a way to pass flags to makeinfo
58 * test `make clean' with subdir-objects
60 * Test nodist_SOURCES with lex, yacc, etc.
62 * Support subdir-objects with fortran
64 * Allow creation of Java .zip/.jar files in natural way
65 If you are building a compiled Java library, then the .zip/.jar
66 ought to be made automatically.
68 * Run automake before libtool. It will report an error but
69 still won't put the file into the disty. This is wrong.
70 From Mark H Wilkinson <mhw@kremvax.demon.co.uk>
72 * CFLAGS only defined if C source seen
73 but really it should be a configure variable, shouldn't it?
74 There are other examples of this
76 * in gnu/gnits mode, give error if Makefile.am overrides a user
79 * If we see `foo.o' in LIBOBJS, and we've seen AC_OBJEXT, then complain.
81 * using "include" inside a conditional doesn't currently
84 * examine possibility of using any character in a macro name
85 and rewriting names automatically. this means we must rewrite
86 all references as well.
88 * AM_CONFIG_HEADER might generate the wrong stamp file names
89 when given multiple headers. Write a test.
91 * Currently don't correctly handle multiple inputs to a config header.
93 * header stamp files still in wrong dirs.
94 stamp-h.in must be in dir with h.in file
95 stamp-h must be in dir with output file
101 ... this ought to work. The fix is probably complicated
103 * `distcheck' and `dist' should depend on `all'
105 * Add code to generate foo-config script like gnome, gtk
107 * `DEFS += foo' won't work.
108 That's because DEFS is defined in header-vars.am, which is read
109 after the user's Makefile.am.
110 This will be a problem for any macro defined internally
111 [ fixing this will probably fix the nasty `exeext redefines
112 foo_PROGRAMS' hack that is in there right now ]
113 In some cases this is unfixable. In these cases we should give
114 an error. This can be done by keeping track of whether a macro
115 was assigned with `=' or `+=', and having define_variable give
116 an error if the variable exists and was defined with +=.
118 * document user namespace for macro/target names
119 adopt some conventions and use uniformly
120 [ this is a good thing for the rewrite ]
122 * make distcheck uses directories like `=build'.
123 Some (very rare) POSIX systems don't support `=' in filenames.
124 If this ever becomes a problem, fix it
126 * distclean must remove config.status
127 can't this cause problems for maintainer-clean?
128 shouldn't maintainer-clean print the message before running
129 any part of the make? (just to slow things down long enough
130 for the user to stop it)
131 (maybe doesn't matter since people who even know about
132 maintainer-clean already have a clue)
134 * There are probably more bugs in variable_conditions_sub along
135 the lines of the one that caused cond4.test to fail.
137 * give user more control over -I flags
138 in particular document a way to override the defaults
140 * reintroduce AM_FUNC_FNMATCH which sets LIBOBJS
141 Then have automake know about fnmatch.h.
142 [ probably should wait for autoconf to get right functionality ]
144 * Add a conditional for dependency tracking
145 (what to name it is the biggest problem here)
146 (because we want it to flag dist/no-dist -- not just deps)
147 [ this might not really be correctly doable.
148 instead we need a compile-time conditional for this
151 * Allow per-object cflags:
154 * per-object compiler flags do not apply to libobjs
155 -> give error in this case
156 * At the same time, allow sources in subdirs:
158 This requires `mkdir x' at build time
159 [ both of these require per-file rules, and not pattern rules ]
160 [ use user-written suffix rules to generate the per-file rules in
161 an automatic way -- this would be mucho cool ]
163 * Allow for multiple translations of a texinfo file:
164 LL_info_TEXINFOS = ...
165 will put info files for language LL into $(infodir)/LL.
167 * If you suppress an internal variable by specifying a variable
168 in a Makefile.am, but the variable is conditional, then automake
169 should generate the internal variable conditionally.
170 You have to scan the map of all conditions and fill in the holes here
172 * Every program foo has FOOFLAGS right now.
173 It should also have AM_FOOFLAGS, which can be set in Makefile.am.
174 DONE: but needs to be documented
176 * Should be able to update files that would be installed with -a
179 * "make diff" capability
180 look at gcc's Makefile.in to see what to do
181 or look at maint program
183 * Karl wants to be able to set LIBS and LDFLAGS at build time, like CFLAGS
184 maybe we need something more general?
186 * add $(srcdir)/ before some dependencies?
188 * define LINK if a program is mentioned, even if no C sources appear
190 * BUILT_SOURCES should not be distributed, even when they appear in
191 another _SOURCES line. [? or maybe just leave this up to the
192 to-be-defined generic distribution method ]
193 must completely revisit the entire BUILT_SOURCES idea
195 * in --cygnus, clean-info not generated at top level
197 * what if an element of a scanned variable looks like
199 or some other arbitrary thing?
200 right now we try to cope, but not very well
202 * if `interlock' exists, that should be an error (?)
203 should also warn about using new ylwrap and not old one
204 only do this when looking for ylwrap
206 ** make sure every variable that is used is also defined
208 * make sure `missing' defines are generated
209 * if no AM_INIT_AUTOMAKE, then don't handle `missing' stuff.
211 * missing should handle install -d and rmdir -p (for uninstall)
213 * a couple ways to be smarter:
214 - notice when a .c file is a target somewhere, and auto-add it to
216 - notice a target of the form `.x.y:' and assume it is a suffix rule
219 * NORMAL_INSTALL / NORMAL_UNINSTALL -vs- recursive rules
220 [ requires changes to the standard ]
222 * cross-compilation support:
223 programs built and used by the build process need to be
224 built for CC_FOR_BUILD
225 introduce a new variable for this
227 * if foo.y is a source, foo.h isn't auto-distributed?
229 * copyrights on m4 files, aclocal output
231 * is there a way to add a directory and then have "make" do all the
234 * put standards.texi into distribution
237 * should not put texiname_TEXINFOS into distribution
238 should rename this macro anyway, to foo_texi_DEPENDENCIES
240 * *all* installed scripts should support --version, --help
242 For now I guess I'll just have automake give an error if it encounters
243 non-C source in a libtool library specification.
245 * must split $obj into two parts: one for libtool and one for
246 deansification. Otherwise .S files will be deansified!
248 * ansi2knr must currently appear in a directory that has some source
250 * if program has the same name as a target, do something sensible:
251 - if the target is internal, rename it
252 - if the target is mandated (eg, "info"), tell the user
253 consider auto-modifying the program name to work around this
255 * should separate actual options from strictness levels
256 strictness should only cover requirements
257 You should be able to pick and choose options
259 should clean up texinfos.am; one rule is repeated 3 times, but
262 should always use perl -w
264 rewrite in guile (RMS request)
265 at the same time, consider adding a GUI
266 could use the same parsing code for the GUI and the standalone version
267 that means figuring out a better representation of internal state
268 [ that's easy -- anything is better than what we have now ]
270 having just one Makefile for a project would give a big speed increase
271 for a project with many directories, eg glibc. ideally (?) you'd
272 still be able to have a Makefile.am in each directory somehow; this
273 might make editing conceptually easier.
275 * finish up TAGS work
277 * put parser.h into distribution if "yacc -d" is used
279 * only remove libtool at top level?
281 * clean up source directory by moving stuff into subdirs
283 * consider adding pkglibexecdir, maybe others?
284 requests for pkg-dirs with version included
286 Avoid loops when installing; instead unroll them in automake
289 * completely handle multi-":" mode for AC_CONFIG_HEADER
290 * Scan multiple input files when Makefile is generated?
291 This would provide flexibility for large projects; subsumes
292 the "Makefile.tmpl" idea
294 [ can't do this. must explain why in manual.
295 basically, solving all the problems is too hard
296 like: how to remove redundancies between generated .in files
297 instead should implement `include' directive for Makefile.am ]
298 * for multi-":" mode and AC_OUTPUT, it might be good to pick the
299 first input file that has a corresponding .am file.
301 Some long-term projects:
302 * if $(FOO) is used somewhere, ensure FOO is defined, either by
303 user or by automake if possible
305 [ include, += support ]
306 * even better would be allowing targets in different included
307 fragments to be merged. e.g., `install-local'.
309 consider putting all check-* targets onto @check?
310 To support --help/--version checking?
312 take diff-n-query code from libit
315 Per> 1) Being able to build a set of non-source programs
316 Per> from source programs, without necessarily linking them together.
317 Per> I.e. one should be able to say something like:
318 Per> dummy_SOURCES=foo.c bar.c
319 Per> and automake should realize that it needs to build foo.o and bar.o.
320 Per> 2) Being intelligent about new kinds of suffixes.
322 Per> SUFFIXES = .class .java
323 Per> and a suffix rule of the form:
325 Per> then it should be able to realize it can build .class files from
326 Per> .java files, and thus be able to generate a list of
327 Per> .class files from a list of .java source files.
329 !! Must fix require_file stuff. It is really gross, and I don't
330 understand it any more.
332 * error messages should print ``[info blah blah]'' command when a
333 certain part of the standards apply. saw idea in message from
334 Craig Burley. wouldn't it be really cool if compile-mode in Emacs
335 understood this convention, and you could click on such text to
336 go to the appropriate info page? [ I think you can ]
338 Jim's idea: should look for @setfilename and warn if filenames too long
341 ** many requests for a way to omit a file from the distribution.
342 Should be done like `!foo' or `~foo' in _SOURCES, etc.
343 Such files should be removed explicitly after the copy step!
344 Doing this requires rewriting macros before generating Makefile.in.
346 from joerg-martin schwarz:
347 -- If Makefile.am contains $(CC), $(COMPILE), $(YLWRAP), ....
348 in an explicitly written rule, you should emit the corresponding
349 Makefile variables automatically.
351 Configuring in the large:
352 * allow hierarchy of dirs to share one aclocal.m4
355 consider printing full file name of Makefile.am or configure.in when
356 giving error. This would help for very large trees with many
359 From the GNU Standards. These things could be checked, and probably
361 * Make sure that the directory into which the distribution unpacks (as
362 well as any subdirectories) are all world-writable (octal mode 777).
363 * Make sure that no file name in the distribution is more than 14
365 * Don't include any symbolic links in the distribution itself.
367 * Make sure that all the files in the distribution are world-readable.
368 ** also, check --help output and --version output. Idea from François
369 * standards no longer prohibit ANSI C. What does this imply
370 for the de-ansi-fication feature? [ must keep it -- some users rely on it ]
372 should be able to determine what is built by looking at rules (and
373 configure.in). Then built man pages (eg) could automatically be
374 omitted from the distribution.
376 Idea from Joerg-Martin Schwarz: allow passing different -D flags to
377 different compiles. This can be done, but with the restriction that a
378 .c cannot appear in 2 different "objects" (programs/libraries)
379 compiled with different -D options (because -c and -o do not always
380 work together and parallel makes must work). This could be
381 implemented by noticing whenever a ".o" target with no rules is being
382 emitted, and adding the appropriate compilation rule as appropriate.
383 This should work with targets from Makefile.am as well as from .P
384 files, which means rewriting so that the Makefile.am contents aren't
385 copied into the output immediately.
386 [ this could be probably done more directly by examining the sources
387 as we scan Makefile.am ]
389 Henrik Frystyk Nielsen says:
390 Henrik> 4) Flags like --include-deps are lost when you make changes to
391 Henrik> Makefile.am files and automake is run automatically. It would
392 Henrik> be nice to keep these flags as I now have to redo everything
394 ... what about other options here too?
396 Think about: maybe "make check" should just bomb if error occurs?
397 Then user must use "make -k check". This is probably more natural.
399 Consider: "cvs" option adds some cvs-specific rules?
401 Right now, targets generated internally (eg "install") are not
402 overridable by user code. This should probably be possible, even
403 though it isn't very important. This could be done by generating all
404 internal rules via a function call instead of just appending to
406 [ this will be harder to implement when scanning a rule like all-recursive
409 * Should be a way to have "nobuild_PROGRAMS" which aren't even built,
410 but which could be by running the magic make command.
413 * Must rewrite am_install_var. Should break into multiple functions.
414 This will allow the callers to be a little smarter.
415 * Rewrite clean targets.
416 * Must rewrite error handling code. Right now it is a real mess
417 Should fix up require_file junk at the same time
419 djm wants ``LINKS'' variable; list of things to link together after
420 install. In BSD environment, use:
421 LINKS = from1 to1 from2 to2 ...
423 Need way to say there are no suffixes in a Makefile (Franc,ois'
424 "override" idea suffices here)
426 Check to make sure various scripts are executable (IE when looking for
429 Use recode in dist target when MAINT_CHARSET specified. Read caveats
430 in automake.in before doing this. Note the same problem used to apply
431 to the no-dependencies option; maybe it still should? Note also that
432 each Makefile.am must be rewritten at "make dist" time if
433 MAINT_CHARSET and DIST_CHARSET are not identical. NOTE: gettext must
434 arrange for all .po files not to be recoded. In the long term this
435 might be a problem (consider when some systems use Unicode but the
437 MAINT_CHARSET *must* be local to each Makefile.am, to enable
438 merged distributions.
439 DIST_CHARSET must be passed down to subdir makes during a "make dist"
441 Handle dist-zoo. Generally add more DOS support. Maybe run "doschk"
442 (why isn't this merged with "pathchk"?) when doing a dist. Do
443 whatever else François says here...
445 Add support for html via an option. Use texi2html. Use
446 "html_TEXINFOS", and htmldir = .../html. Include html files in
447 distribution. Also allow "html_DATA", for raw .html files.
448 [ when will texinfo directly support html? ]
450 uninstall and pkg-dirs should rm -rf the dir.
452 a potential bug: configure puts "blah.o" into LIBOBJS, thus implying
453 these files can't be de-ansified. Not a problem?
454 [ fix by using ansi2knr wrapper program ]
456 In general most .am files should be merged into automake. For
457 instance all the "clean" targets could be merged by keeping lists of
458 things to be removed. This would be a lot nicer looking. Note that
459 the install targets probably should not be merged; it is sometimes
460 useful to only install a small part.
463 * Order rules sensibly
464 * Ensure every line has a purpose. Omit unused stuff
465 * Eliminate extraneous rules when possible (eg 'install-am' stuff)
466 * Make sure vertical spacing is correct
467 Omit program transform vars from header if no program installed. This
468 is currently pretty hard to do. (But with beautification code it
469 would probably be easy)
472 * It would be nice to automatically support using bison's better features
473 to rename the output files. This requires autoconf support
474 * Consider supporting syntax from autoconf "derived:source", eg:
476 for yacc and lex source
477 * what if you use flex and the option to avoid -lfl?
480 Multi-language support:
481 * should have mapping of file extensions to languages
482 * should automatically handle the linking issue (special-case C++)
483 * must get compile rules for various languages; FORTRAN probably
484 most important unimplemented language
485 This should be integrated in some way with Per's idea.
486 Eg .f.o rules should be recognized & auto-handled in _SOURCES
487 That way any random language can be treated with C/C++ on a first-class
490 It might be cool to generate .texi dependencies by grepping for
491 @include. (If done, it should be done the same way C dependencies are
494 It would be good to check some parts of GNU standards. Already check
495 for install-sh and mkinstalldirs. What else is required to be in
496 package by GNU standards or by automake?
497 Some things for --strictness=gnits:
498 * "cd $(foo); something" is an error in a rule. Should be:
499 "cd $(foo) && something"
500 * Look for 'ln -s' and warn about using $(LN) and AC_PROG_LN_S
501 * Look for $(LN) and require AC_PROG_LN_S
503 Auto-distribute "ChangeLog.[0-9]+"? "ChangeLog.[a-z]+"?
505 Internationalize. [ gettext doesn't have the necessary machinery yet ]
506 am_error should use printf-style arguments (for eventual gettext scheme)
508 François says the ordering of files in a distribution should be as follows:
512 I agree, but I don't see how to implement this yet.
513 It might be easier if "derived files" is limited to those that
514 Automake itself knows about, eg output of yacc.
516 Check all source files to make sure that FSF address is up-to-date.
517 --gnits or --gnu only.
519 Merge each -vars.am file with corresponding ".am" file. Can do this
520 because of changes to &file_contents.
522 Should libexec programs have the name transform done on them?
524 Order the output rules sensibly, so FOO_SOURCES and FOO_OBJECTS are
525 together and rules are in the usual order.
527 Make the output minimal: only output definitions for variables that
531 David> To avoid comments like the one about subdirs getting buried in
532 David> the middle of a Makefile.in, how about pushing comments that
533 David> start with ### to the top of the Makefile.in (in order)? Sort
534 David> of like how Autoconf uses diversions to force initialization
535 David> code to the top of configure.
538 Karl> 2) Your Makefile variable names are generally uppercase, but GNU
539 Karl> generally uses lowercase. Not that it matters :-).
541 ================================================================
545 probably should put each group of m4 files into a subdir owned by the
546 containing application.
548 ================================================================
554 how to use the generated makefiles
557 - NORMAL_INSTALL junk
559 what goes in AC_CONFIG_AUX_DIR
561 multi-":" mode in AC_OUTPUT -- automake only looks at the first file
562 also a note on how a .am file is found in this case
564 rationale for avoiding
565 make CFLAGS="$CFLAGS" ...
568 a package that installs its own aclocal macros
570 write example of using automake with dejagnu
571 follow calc example in dejagnu docs
573 document which variables are actually scanned and which are not.
575 Document customary ordering of Makefile.am. From François.
577 Should include extended version of diagram from Autoconf (suggested by
580 Make a definition of the term "source"
582 document how to use Automake with CVS. Idea from Mark Galassi. Also
583 include Greg Woods' more sophisticated "cvs-dist" target.
585 document rebuilding configure. CONFIGURE_DEPENDENCIES
586 CONFIG_STATUS_DEPENDENCIES
588 -- must document all variables that are supposed
589 to be public knowledge
591 must document the targets required for integration with
592 non-automake-using subdirs
594 document the "make SHELL='/bin/sh -x'" trick for debugging
596 section on relationship to GNU make. include notes on parallel makes
600 move discussion of cygwin32, etags, mkid under other gnu tools
604 ================================================================
606 Things to do for gcc:
608 Regularize dependency generation. Add new flags:
610 -MH Generate a dummy dependency for each header file mentioned.
614 Set name of output file
616 Then automake can use -MD -MH -MT 'foo.o foo.lo' -MF .deps/...
618 ================================================================
620 Things to do for autoconf:
622 * patch autoreconf to run automake and aclocal. I've done this but it is
623 not really available. It can't be made available until automake
624 is officially released
626 ================================================================
630 * Should support standalone library along with subdir library in same
631 Makefile.am. Maybe: turn off "standalone" mode if library's Makefile.am
632 is not only one specd? [ add an option for this ]
634 ================================================================
638 Would it be useful to integrate in some way with the Debian package
639 building utility? Must check. maybe it would be possible to deal
640 with all the different package utilities somehow. Lately I've been
641 hearing good things about the RedHat packaging utilities. Why are
642 there so many of these? Are they fun to write or something?
643 The RedHat package utility is called RPM; see
644 ftp://ftp.redhat.com/pub/code/rpm
645 It actually has problems, like no configure script and no documentation.
647 For Cygnus it would probably be good to be able to handle the native
648 package utility on each platform. There are probably 3 or 4 of these
649 (sysv, solaris?, aix?)
651 tcl/unix/Makefile.in has some code to generate a Solaris package.
653 Automake probably can't do all of this on its own. A new tool might
656 I have some notes from a Debian developer on how the integration
659 ================================================================
661 A tool to guess what the local Makefile.am should look like:
662 (see Gord's Maint program!)
664 * Probably integrate with autoscan
665 * Use various simple rules to determine what to do:
666 * get name of top directory, sans version info
667 * search for .c files with 'main' in them
668 * if in main.c, use directory name for program
669 * if in more than one, generate multiple programs
670 * if not found, generate a library named after directory
671 * order subdir searches correctly: lib first, src last
672 * assume 'testsuite' dir means we are using dejagnu
673 * maybe be smart about reading existing Makefile.am, so tool
674 can be run for incremental changes? You could imagine:
677 autoproject --incremental
679 ================================================================
681 Stuff NOT to do, and why:
683 consider auto-including any file that matches "*.in".
684 [ no: po/Makefile.in shouldn't be included ]
686 must look at mkid to see how it works (for subdir usage)
687 [ right now, it doesn't. i don't see a simple fix right now ]
689 if configure.in not found, move up a directory and try again? This
690 could eliminate a common source of problems.
691 [ this is just a bad idea ]
693 * scripts are installed in $exec_prefix/bin, not $prefix/bin
695 [ the consensus on Gnits is that this isn't required.
696 doubters can work around it anyway ]
698 * make the auto-dep code crash if GNU make not in use?
699 (doesn't it already?)
701 Looked at a program called 'ezmake', which seems to do something
702 similar. The only idea there that is possibly worth stealing is using
703 globs in definitions. Also has negations. Eg in a directory with
704 files a.c, b.c and c.c, the line:
705 foo_SOURCES = *.c ~c.c
706 would be equivalent to:
707 foo_SOURCES = a.c b.c
708 Is this worth implementing?
709 [ No... it is more reliable to spell everything out. ]
711 Scan source directories and warn about missing files, eg .c/.h files
712 that aren't mentioned?
713 [ distcheck makes this less useful ]
716 - how to install file with a space in its name?
717 [ don't bother with this -- make is just too losing ]