1 we can't seem to AC_SUBST(pkgdatadir)
2 the version from header-vars overrides
5 check should depend on all
8 refactor handle_source_transform and handle_single_transform_list
9 we suffer combinatorial explosion here, but there's no need to
10 we could transform source variables one at a time
12 the new YFLAGS code doesn't correctly handle srcdir
14 allow foo_NAME to rename an object (library or program)
17 remove _LTLIBRARIES and just use _LIBRARIES
18 then use this for zip/jar as well
20 consider `sub am_unique_error', which only prints a given error
21 message once. Use eg in lang_c_rewrite for ansi2knr error
24 investigate problems with conditionally defined libraries
26 add an error if the user makefile.am violates our
29 have 'make check' print tests which are skipped
31 we need a document describing automake from the end user's point of view
32 eg describe INSTALL_HEADER there, among other things
38 > DISTCLEAN -test -z "$(DISTCLEANFILES)" || rm -f $(DISTCLEANFILES)
40 > maintainer-clean-generic:
41 > +## FIXME: shouldn't we really print these messages before running
42 > +## the dependencies?
43 > + @echo "This command is intended for maintainers to use"
44 > + @echo "it deletes files that may require special tools to rebuild."
48 > I'd like to eventually fix the FIXME comment by having
49 > maintainer-clean look like:
55 > We're left with the question of whether we should repeat them in every
60 > Hmm... Interesting. It must have been a side effect of the enabling
61 > of forced `relink' on GNU/Linux/x86. Anyway, on platforms that
62 > actually require relinking, this problem remains, and I see no way to
63 > overcome it other than arranging for automake to install libraries
64 > before executables, as you suggest. This shouldn't be a big problem,
67 > A bigger problem could show up if two libraries in the same directory,
68 > one dependent on the other, are installed concurrently. If relinking
69 > is needed for the dependent library, we have a problem. It appears to
70 > me that user will have to live without `make -j install', in this
74 > Here's an Automake patch and changelog entry allow make -j install on
75 > such degenerate systems (and Linux with buggy libtool <g>)
77 > If you install to locations other that bin_ and lib_ then a larger fix
78 > is necessary, but this should fix the 90% case.
80 * think about how per-object flags should work. in particular:
81 * how should they be specified?
82 using the object name is confusing when .lo/.obj in use
83 however, the object name provides a nice interaction with
85 * how should they interact with per-executable flags?
86 [ this is probably a feature in search of a problem ]
88 * cross-compilation support:
89 programs built and used by the build process need to be
90 built for CC_FOR_BUILD
91 introduce a new prefxi for this, e.g. `build_PROGRAMS'
92 [ we can do this in an automatic way I think.
93 unfortunately it isn't that useful until autoconf has support
94 for this sort of thing as well ]
96 * distcheck should make sure that each file that uses _() is
99 # Verify that all source files using _() are listed in po/POTFILES.in.
101 grep -E -v '^(#|$$)' po/POTFILES.in | sort > $@-1
102 grep -E -l '\b_\(' lib/*.c src/*.c | sort > $@-2
106 * one performance enhancement would be to have autoconf write
107 a single file containing all the macro assignments.
108 then read this file via `include'
109 unfortunately this can't be done because of conditionals
110 -- but it could be made to work if we also changed:
111 * automake to rewrite @FOO@ to $(FOO), and
112 * the implementation of conditionals to rely on some new
115 * support prog_LIBS as override for LIBS
117 * Scan configure.in using traces as autoheader does.
118 This will be much more reliable.
120 * Test subdir-objects option with yacc, lex, ansi2knr
121 Our locking scheme won't prevent a parallel make from losing
122 if there are two `bar.o' files and the timing is just right
123 This only happens with parallel make and no-`-c -o' compiler,
124 so it probably isn't very important
125 `-c -o' when doing libtool
126 try to find a losing compiler and see if it really works.
127 (actually: hack config.cache and do it)
130 ** We're using `$<' in explicit rules when using per-exe flags
131 ** per-exe flags don't work for CPPFLAGS/YFLAGS/LFLAGS. Fix.
132 ** LIBOBJS shouldn't be used when there are per-exe flags (?)
134 * Test nodist_SOURCES with lex, yacc, etc.
136 * Support subdir-objects with fortran
138 * Allow creation of Java .zip/.jar files in natural way
139 If you are building a compiled Java library, then the .zip/.jar
140 ought to be made automatically.
142 * Run automake before libtool. It will report an error but
143 still won't put the file into the disty. This is wrong.
144 From Mark H Wilkinson <mhw@kremvax.demon.co.uk>
146 * CFLAGS only defined if C source seen
147 but really it should be a configure variable, shouldn't it?
148 There are other examples of this
149 [ moving to autoconf --trace ought to fix this ]
151 * in gnu/gnits mode, give error if Makefile.am overrides a user
152 variable like CFLAGS.
153 [ this is low priority because the package author can always
154 circumvent our check by redefining in configure.in
155 plus it is probably better to encourage good behavior than to
158 * If we see `foo.o' in LIBOBJS, and we've seen AC_OBJEXT, then complain.
159 [ how will we know that? it is better to handle this automatically
160 via an autoconf hook ]
162 * examine possibility of using any character in a macro name
163 and rewriting names automatically. this means we must rewrite
164 all references as well.
165 [ this is a 2.0-style feature ]
167 * AM_CONFIG_HEADER might generate the wrong stamp file names
168 when given multiple headers. Write a test.
170 * Currently don't correctly handle multiple inputs to a config header.
171 [ this should no matter in the future as acconfig.h and so on are
172 obsoleted by the AH series of macros.]
174 * header stamp files still in wrong dirs.
175 stamp-h.in must be in dir with h.in file
176 stamp-h must be in dir with output file
178 * conditionals and macros
179 Our current scheme cause combinatoric explosion.
181 In fact, to be honest, I no longer understand very well why we perform
182 such a closure. I mean, as is, Automake transforms (this is
185 | bin_PROGRAMS = targ
205 | targ_SOURCES = $(SONE) $(STWO) $(STHREE)
209 | @ONE_FALSE@@THREE_FALSE@@TWO_TRUE@am_targ_OBJECTS = two.$(OBJEXT)
210 | @ONE_FALSE@@THREE_FALSE@@TWO_FALSE@am_targ_OBJECTS =
211 | @ONE_FALSE@@THREE_TRUE@@TWO_TRUE@am_targ_OBJECTS = two.$(OBJEXT) \
212 | @ONE_FALSE@@THREE_TRUE@@TWO_TRUE@ three.$(OBJEXT)
213 | @ONE_FALSE@@THREE_TRUE@@TWO_FALSE@am_targ_OBJECTS = three.$(OBJEXT)
214 | @ONE_TRUE@@THREE_FALSE@@TWO_TRUE@am_targ_OBJECTS = one.$(OBJEXT) \
215 | @ONE_TRUE@@THREE_FALSE@@TWO_TRUE@ two.$(OBJEXT)
216 | @ONE_TRUE@@THREE_FALSE@@TWO_FALSE@am_targ_OBJECTS = one.$(OBJEXT)
217 | @ONE_TRUE@@THREE_TRUE@@TWO_TRUE@am_targ_OBJECTS = one.$(OBJEXT) \
218 | @ONE_TRUE@@THREE_TRUE@@TWO_TRUE@ two.$(OBJEXT) three.$(OBJEXT)
219 | @ONE_TRUE@@THREE_TRUE@@TWO_FALSE@am_targ_OBJECTS = one.$(OBJEXT) \
220 | @ONE_TRUE@@THREE_TRUE@@TWO_FALSE@ three.$(OBJEXT)
222 why don't we just output
224 | @ONE_TRUE@am_SONE_OBJECTS = one.$(OBJEXT)
225 | @ONE_FALSE@am_SONE_OBJECTS =
227 | @TWO_TRUE@am_STWO_OBJECTS = two.$(OBJEXT)
228 | @TWO_FALSE@am_STWO_OBJECTS =
230 | @THREE_TRUE@am_STHREE_OBJECTS = three.$(OBJEXT)
231 | @THREE_FALSE@am_STHREE_OBJECTS =
233 | am_targ_OBJECTS = $(am_SONE_OBJECTS) $(am_STWO_OBJECTS) $(am_STHREE_OBJECTS)
235 which means also, why do we look for the closure of PROGRAMS, instead
236 of just adding $(EXEEXT) to all its components and sub components
237 (i.e., inside sub vars such as $(SONE) above being a sub var of
241 Aaaaaaaaaaah! I think I know... Must be because of `+='.
243 Hm... No. Indeed we transform
252 | @BAR_TRUE@FOO = foo bar
253 | @BAR_FALSE@FOO = foo
255 but this seems good to me too?
257 | FOO = foo $(BAR_FOO)
258 | @BAR_TRUE@BAR_FOO = bar
259 | @BAR_FALSE@BAR_FOO =
266 ... this ought to work. The fix is probably complicated, but might
267 come for free when we rewrite the handling of conditionals.
269 * `distcheck' and `dist' should depend on `all'
271 * Add code to generate foo-config script like gnome, gtk
273 * `DEFS += foo' won't work.
274 That's because DEFS is defined in header-vars.am, which is read
275 after the user's Makefile.am.
276 This will be a problem for any macro defined internally
277 [ fixing this will probably fix the nasty `exeext redefines
278 foo_PROGRAMS' hack that is in there right now ]
279 [ we currently give an error when this occurs, so this is very low
282 * document user namespace for macro/target names
283 adopt some conventions and use uniformly
284 [ this is a good thing for the rewrite ]
286 * make distcheck uses directories like `=build'.
287 Some (very rare) POSIX systems don't support `=' in filenames.
288 If this ever becomes a problem, fix it
290 * distclean must remove config.status
291 can't this cause problems for maintainer-clean?
292 shouldn't maintainer-clean print the message before running
293 any part of the make? (just to slow things down long enough
294 for the user to stop it)
295 (maybe doesn't matter since people who even know about
296 maintainer-clean already have a clue)
298 * reintroduce AM_FUNC_FNMATCH which sets LIBOBJS
299 Then have automake know about fnmatch.h.
300 [ probably should wait for autoconf to get right functionality ]
302 * "make diff" capability
303 look at gcc's Makefile.in to see what to do
304 or look at maint program
306 * in --cygnus, clean-info not generated at top level
308 * what if an element of a scanned variable looks like
310 or some other arbitrary thing?
311 right now we try to cope, but not very well
312 [ this is only of theoretical interest for now ]
314 * make sure every variable that is used is also defined
315 [ we don't really look at variable uses in detail.
318 * make sure `missing' defines are generated
320 * missing should handle install -d and rmdir -p (for uninstall)
322 * notice when a .c file is a target somewhere, and auto-add it to
325 * NORMAL_INSTALL / NORMAL_UNINSTALL -vs- recursive rules
326 [ requires changes to the standard ]
328 * copyrights on m4 files, aclocal output
330 * should not put texiname_TEXINFOS into distribution
331 should rename this macro anyway, to foo_texi_DEPENDENCIES
333 * *all* installed scripts should support --version, --help
335 * For now I guess I'll just have automake give an error if it encounters
336 non-C source in a libtool library specification.
338 * must split $obj into two parts: one for libtool and one for
339 deansification. Otherwise .S files will be deansified!
341 * ansi2knr must currently appear in a directory that has some source
343 * if program has the same name as a target, do something sensible:
344 - if the target is internal, rename it
345 - if the target is mandated (eg, "info"), tell the user
346 consider auto-modifying the program name to work around this
348 * should separate actual options from strictness levels
349 strictness should only cover requirements
350 You should be able to pick and choose options
352 * rewrite in guile (RMS request)
353 at the same time, consider adding a GUI
354 could use the same parsing code for the GUI and the standalone version
355 that means figuring out a better representation of internal state
356 [ that's easy -- anything is better than what we have now ]
358 having just one Makefile for a project would give a big speed increase
359 for a project with many directories, eg glibc. ideally (?) you'd
360 still be able to have a Makefile.am in each directory somehow; this
361 might make editing conceptually easier.
363 * finish up TAGS work
365 * only remove libtool at top level?
367 * clean up source directory by moving stuff into subdirs
369 * consider adding pkglibexecdir, maybe others?
370 requests for pkg-dirs with version included
372 Avoid loops when installing; instead unroll them in automake
375 * completely handle multi-":" mode for AC_CONFIG_HEADER
376 * Scan multiple input files when Makefile is generated?
377 This would provide flexibility for large projects; subsumes
378 the "Makefile.tmpl" idea
380 [ can't do this. must explain why in manual.
381 basically, solving all the problems is too hard
382 like: how to remove redundancies between generated .in files
383 instead should implement `include' directive for Makefile.am ]
384 * for multi-":" mode and AC_OUTPUT, it might be good to pick the
385 first input file that has a corresponding .am file.
387 Some long-term projects:
388 * if $(FOO) is used somewhere, ensure FOO is defined, either by
389 user or by automake if possible
391 [ include, += support ]
392 * even better would be allowing targets in different included
393 fragments to be merged. e.g., `install-local'.
395 consider putting all check-* targets onto @check?
396 To support --help/--version checking?
398 take diff-n-query code from libit
401 Per> 1) Being able to build a set of non-source programs
402 Per> from source programs, without necessarily linking them together.
403 Per> I.e. one should be able to say something like:
404 Per> dummy_SOURCES=foo.c bar.c
405 Per> and automake should realize that it needs to build foo.o and bar.o.
406 Per> 2) Being intelligent about new kinds of suffixes.
408 Per> SUFFIXES = .class .java
409 Per> and a suffix rule of the form:
411 Per> then it should be able to realize it can build .class files from
412 Per> .java files, and thus be able to generate a list of
413 Per> .class files from a list of .java source files.
415 !! Must fix require_file stuff. It is really gross, and I don't
416 understand it any more.
418 Jim's idea: should look for @setfilename and warn if filenames too long
421 from joerg-martin schwarz:
422 -- If Makefile.am contains $(CC), $(COMPILE), $(YLWRAP), ....
423 in an explicitly written rule, you should emit the corresponding
424 Makefile variables automatically.
426 Configuring in the large:
427 * allow hierarchy of dirs to share one aclocal.m4
430 consider printing full file name of Makefile.am or configure.in when
431 giving error. This would help for very large trees with many
434 From the GNU Standards. These things could be checked, and probably
436 * Make sure that the directory into which the distribution unpacks (as
437 well as any subdirectories) are all world-writable (octal mode 777).
438 * Make sure that no file name in the distribution is more than 14
440 * Don't include any symbolic links in the distribution itself.
442 * Make sure that all the files in the distribution are world-readable.
443 ** also, check --help output and --version output. Idea from François
444 * standards no longer prohibit ANSI C. What does this imply
445 for the de-ansi-fication feature? [ must keep it -- some users rely on it ]
447 should be able to determine what is built by looking at rules (and
448 configure.in). Then built man pages (eg) could automatically be
449 omitted from the distribution.
451 Henrik Frystyk Nielsen says:
452 Henrik> 4) Flags like --include-deps are lost when you make changes to
453 Henrik> Makefile.am files and automake is run automatically. It would
454 Henrik> be nice to keep these flags as I now have to redo everything
456 ... what about other options here too?
458 Think about: maybe "make check" should just bomb if error occurs?
459 Then user must use "make -k check". This is probably more natural.
461 Consider: "cvs" option adds some cvs-specific rules?
463 Right now, targets generated internally (eg "install") are not
464 overridable by user code. This should probably be possible, even
465 though it isn't very important. This could be done by generating all
466 internal rules via a function call instead of just appending to
468 [ this will be harder to implement when scanning a rule like all-recursive
471 * Should be a way to have "nobuild_PROGRAMS" which aren't even built,
472 but which could be by running the magic make command.
475 * Must rewrite am_install_var. Should break into multiple functions.
476 This will allow the callers to be a little smarter.
477 * Rewrite clean targets.
478 * Must rewrite error handling code. Right now it is a real mess
479 Should fix up require_file junk at the same time
481 djm wants ``LINKS'' variable; list of things to link together after
482 install. In BSD environment, use:
483 LINKS = from1 to1 from2 to2 ...
485 Need way to say there are no suffixes in a Makefile (Franc,ois'
486 "override" idea suffices here)
488 Check to make sure various scripts are executable (IE when looking for
491 Handle dist-zoo. Generally add more DOS support. Maybe run "doschk"
492 (why isn't this merged with "pathchk"?) when doing a dist. Do
493 whatever else François says here...
495 Add support for html via an option. Use texi2html. Use
496 "html_TEXINFOS", and htmldir = .../html. Include html files in
497 distribution. Also allow "html_DATA", for raw .html files.
498 [ when will texinfo directly support html? ]
500 uninstall and pkg-dirs should rm -rf the dir.
502 a potential bug: configure puts "blah.o" into LIBOBJS, thus implying
503 these files can't be de-ansified. Not a problem?
504 [ fix by using ansi2knr wrapper program ]
506 In general most .am files should be merged into automake. For
507 instance all the "clean" targets could be merged by keeping lists of
508 things to be removed. This would be a lot nicer looking. Note that
509 the install targets probably should not be merged; it is sometimes
510 useful to only install a small part.
512 * Clean up the output:
513 ** Order rules sensibly
514 ** Ensure every line has a purpose. Omit unused stuff
515 ** Eliminate extraneous rules when possible (eg 'install-am' stuff)
516 ** Make sure vertical spacing is correct
517 Omit program transform vars from header if no program installed. This
518 is currently pretty hard to do. (But with beautification code it
519 would probably be easy)
522 ** It would be nice to automatically support using bison's better features
523 to rename the output files. This requires autoconf support
524 ** Consider supporting syntax from autoconf "derived:source", eg:
526 for yacc and lex source
527 ** what if you use flex and the option to avoid -lfl?
530 * Multi-language support:
531 ** should have mapping of file extensions to languages
532 ** should automatically handle the linking issue (special-case C++)
533 ** must get compile rules for various languages; FORTRAN probably
534 most important unimplemented language
535 This should be integrated in some way with Per's idea.
536 Eg .f.o rules should be recognized & auto-handled in _SOURCES
537 That way any random language can be treated with C/C++ on a first-class
540 It might be cool to generate .texi dependencies by grepping for
541 @include. (If done, it should be done the same way C dependencies are
544 It would be good to check some parts of GNU standards. Already check
545 for install-sh and mkinstalldirs. What else is required to be in
546 package by GNU standards or by automake?
547 Some things for --strictness=gnits:
548 * "cd $(foo); something" is an error in a rule. Should be:
549 "cd $(foo) && something"
550 * Look for 'ln -s' and warn about using $(LN) and AC_PROG_LN_S
551 * Look for $(LN) and require AC_PROG_LN_S
553 Auto-distribute "ChangeLog.[0-9]+"? "ChangeLog.[a-z]+"?
555 Internationalize. [ gettext doesn't have the necessary machinery yet ]
556 am_error should use printf-style arguments (for eventual gettext scheme)
558 François says the ordering of files in a distribution should be as follows:
562 I agree, but I don't see how to implement this yet.
563 It might be easier if "derived files" is limited to those that
564 Automake itself knows about, eg output of yacc.
566 Check all source files to make sure that FSF address is up-to-date.
567 --gnits or --gnu only.
569 Merge each -vars.am file with corresponding ".am" file. Can do this
570 because of changes to &file_contents.
572 Should libexec programs have the name transform done on them?
574 Order the output rules sensibly, so FOO_SOURCES and FOO_OBJECTS are
575 together and rules are in the usual order.
577 Make the output minimal: only output definitions for variables that
581 David> To avoid comments like the one about subdirs getting buried in
582 David> the middle of a Makefile.in, how about pushing comments that
583 David> start with ### to the top of the Makefile.in (in order)? Sort
584 David> of like how Autoconf uses diversions to force initialization
585 David> code to the top of configure.
588 Karl> 2) Your Makefile variable names are generally uppercase, but GNU
589 Karl> generally uses lowercase. Not that it matters :-).
591 ================================================================
595 probably should put each group of m4 files into a subdir owned by the
596 containing application.
598 ================================================================
604 how to use the generated makefiles
607 - NORMAL_INSTALL junk
609 what goes in AC_CONFIG_AUX_DIR
611 multi-":" mode in AC_OUTPUT -- automake only looks at the first file
612 also a note on how a .am file is found in this case
614 rationale for avoiding
615 make CFLAGS="$CFLAGS" ...
618 a package that installs its own aclocal macros
620 write example of using automake with dejagnu
621 follow calc example in dejagnu docs
623 document which variables are actually scanned and which are not.
625 Document customary ordering of Makefile.am. From François.
627 Should include extended version of diagram from Autoconf (suggested by
630 Make a definition of the term "source"
632 document how to use Automake with CVS. Idea from Mark Galassi. Also
633 include Greg Woods' more sophisticated "cvs-dist" target.
635 document rebuilding configure. CONFIGURE_DEPENDENCIES
636 CONFIG_STATUS_DEPENDENCIES
638 -- must document all variables that are supposed
639 to be public knowledge
641 must document the targets required for integration with
642 non-automake-using subdirs
644 document the "make SHELL='/bin/sh -x'" trick for debugging
646 section on relationship to GNU make. include notes on parallel makes
650 move discussion of cygwin32, etags, mkid under other gnu tools
654 ================================================================
656 Things to do for gcc:
658 Regularize dependency generation. Add new flags:
660 -MH Generate a dummy dependency for each header file mentioned.
664 Set name of output file
666 Then automake can use -MD -MH -MT 'foo.o foo.lo' -MF .deps/...
668 ================================================================
672 * Should support standalone library along with subdir library in same
673 Makefile.am. Maybe: turn off "standalone" mode if library's Makefile.am
674 is not only one specd? [ add an option for this ]
676 ================================================================
680 Would it be useful to integrate in some way with the Debian package
681 building utility? Must check. maybe it would be possible to deal
682 with all the different package utilities somehow. Lately I've been
683 hearing good things about the RedHat packaging utilities. Why are
684 there so many of these? Are they fun to write or something?
685 The RedHat package utility is called RPM; see
686 ftp://ftp.redhat.com/pub/code/rpm
687 It actually has problems, like no configure script and no documentation.
689 For Cygnus it would probably be good to be able to handle the native
690 package utility on each platform. There are probably 3 or 4 of these
691 (sysv, solaris?, aix?)
693 tcl/unix/Makefile.in has some code to generate a Solaris package.
695 Automake probably can't do all of this on its own. A new tool might
698 I have some notes from a Debian developer on how the integration
701 ================================================================
703 A tool to guess what the local Makefile.am should look like:
704 (see Gord's Maint program!)
706 * Probably integrate with autoscan
707 * Use various simple rules to determine what to do:
708 * get name of top directory, sans version info
709 * search for .c files with 'main' in them
710 * if in main.c, use directory name for program
711 * if in more than one, generate multiple programs
712 * if not found, generate a library named after directory
713 * order subdir searches correctly: lib first, src last
714 * assume 'testsuite' dir means we are using dejagnu
715 * maybe be smart about reading existing Makefile.am, so tool
716 can be run for incremental changes? You could imagine:
719 autoproject --incremental
721 ================================================================
723 Stuff NOT to do, and why:
725 consider auto-including any file that matches "*.in".
726 [ no: po/Makefile.in shouldn't be included ]
728 must look at mkid to see how it works (for subdir usage)
729 [ right now, it doesn't. i don't see a simple fix right now ]
731 if configure.in not found, move up a directory and try again? This
732 could eliminate a common source of problems.
733 [ this is just a bad idea ]
735 * scripts are installed in $exec_prefix/bin, not $prefix/bin
737 [ the consensus on Gnits is that this isn't required.
738 doubters can work around it anyway ]
740 Scan source directories and warn about missing files, eg .c/.h files
741 that aren't mentioned?
742 [ distcheck makes this less useful ]
745 - how to install file with a space in its name?
746 [ don't bother with this -- make is just too losing ]
750 Copyright 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001 Free Software
753 This file is part of GNU Automake.
755 GNU Automake is free software; you can redistribute it and/or modify
756 it under the terms of the GNU General Public License as published by
757 the Free Software Foundation; either version 2, or (at your option)
760 GNU Automake is distributed in the hope that it will be useful,
761 but WITHOUT ANY WARRANTY; without even the implied warranty of
762 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
763 GNU General Public License for more details.
765 You should have received a copy of the GNU General Public License
766 along with autoconf; see the file COPYING. If not, write to
767 the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
768 Boston, MA 02111-1307, USA.