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 we need a document describing automake from the end user's point of view
30 eg describe INSTALL_HEADER there, among other things
36 > DISTCLEAN -test -z "$(DISTCLEANFILES)" || rm -f $(DISTCLEANFILES)
38 > maintainer-clean-generic:
39 > +## FIXME: shouldn't we really print these messages before running
40 > +## the dependencies?
41 > + @echo "This command is intended for maintainers to use"
42 > + @echo "it deletes files that may require special tools to rebuild."
46 > I'd like to eventually fix the FIXME comment by having
47 > maintainer-clean look like:
53 > We're left with the question of whether we should repeat them in every
58 > Hmm... Interesting. It must have been a side effect of the enabling
59 > of forced `relink' on GNU/Linux/x86. Anyway, on platforms that
60 > actually require relinking, this problem remains, and I see no way to
61 > overcome it other than arranging for automake to install libraries
62 > before executables, as you suggest. This shouldn't be a big problem,
65 > A bigger problem could show up if two libraries in the same directory,
66 > one dependent on the other, are installed concurrently. If relinking
67 > is needed for the dependent library, we have a problem. It appears to
68 > me that user will have to live without `make -j install', in this
72 > Here's an Automake patch and changelog entry allow make -j install on
73 > such degenerate systems (and Linux with buggy libtool <g>)
75 > If you install to locations other that bin_ and lib_ then a larger fix
76 > is necessary, but this should fix the 90% case.
78 * think about how per-object flags should work. in particular:
79 * how should they be specified?
80 using the object name is confusing when .lo/.obj in use
81 however, the object name provides a nice interaction with
83 * how should they interact with per-executable flags?
84 [ this is probably a feature in search of a problem ]
86 * cross-compilation support:
87 programs built and used by the build process need to be
88 built for CC_FOR_BUILD
89 introduce a new prefxi for this, e.g. `build_PROGRAMS'
90 [ we can do this in an automatic way I think.
91 unfortunately it isn't that useful until autoconf has support
92 for this sort of thing as well ]
94 * distcheck should make sure that each file that uses _() is
97 # Verify that all source files using _() are listed in po/POTFILES.in.
99 grep -E -v '^(#|$$)' po/POTFILES.in | sort > $@-1
100 grep -E -l '\b_\(' lib/*.c src/*.c | sort > $@-2
104 * one performance enhancement would be to have autoconf write
105 a single file containing all the macro assignments.
106 then read this file via `include'
107 unfortunately this can't be done because of conditionals
108 -- but it could be made to work if we also changed:
109 * automake to rewrite @FOO@ to $(FOO), and
110 * the implementation of conditionals to rely on some new
113 * support prog_LIBS as override for LIBS
115 * Scan configure.in using traces as autoheader does.
116 This will be much more reliable.
118 * Test subdir-objects option with yacc, lex, ansi2knr
119 Our locking scheme won't prevent a parallel make from losing
120 if there are two `bar.o' files and the timing is just right
121 This only happens with parallel make and no-`-c -o' compiler,
122 so it probably isn't very important
123 `-c -o' when doing libtool
124 try to find a losing compiler and see if it really works.
125 (actually: hack config.cache and do it)
128 ** We're using `$<' in explicit rules when using per-exe flags
129 ** per-exe flags don't work for CPPFLAGS/YFLAGS/LFLAGS. Fix.
130 ** LIBOBJS shouldn't be used when there are per-exe flags (?)
132 * Test nodist_SOURCES with lex, yacc, etc.
134 * Support subdir-objects with fortran
136 * Allow creation of Java .zip/.jar files in natural way
137 If you are building a compiled Java library, then the .zip/.jar
138 ought to be made automatically.
140 * Run automake before libtool. It will report an error but
141 still won't put the file into the disty. This is wrong.
142 From Mark H Wilkinson <mhw@kremvax.demon.co.uk>
144 * CFLAGS only defined if C source seen
145 but really it should be a configure variable, shouldn't it?
146 There are other examples of this
147 [ moving to autoconf --trace ought to fix this ]
149 * in gnu/gnits mode, give error if Makefile.am overrides a user
150 variable like CFLAGS.
151 [ this is low priority because the package author can always
152 circumvent our check by redefining in configure.in
153 plus it is probably better to encourage good behavior than to
156 * If we see `foo.o' in LIBOBJS, and we've seen AC_OBJEXT, then complain.
157 [ how will we know that? it is better to handle this automatically
158 via an autoconf hook ]
160 * examine possibility of using any character in a macro name
161 and rewriting names automatically. this means we must rewrite
162 all references as well.
163 [ this is a 2.0-style feature ]
165 * AM_CONFIG_HEADER might generate the wrong stamp file names
166 when given multiple headers. Write a test.
168 * Currently don't correctly handle multiple inputs to a config header.
169 [ this should no matter in the future as acconfig.h and so on are
170 obsoleted by the AH series of macros.]
172 * header stamp files still in wrong dirs.
173 stamp-h.in must be in dir with h.in file
174 stamp-h must be in dir with output file
176 * conditionals and macros
177 Our current scheme cause combinatoric explosion.
179 In fact, to be honest, I no longer understand very well why we perform
180 such a closure. I mean, as is, Automake transforms (this is
183 | bin_PROGRAMS = targ
203 | targ_SOURCES = $(SONE) $(STWO) $(STHREE)
207 | @ONE_FALSE@@THREE_FALSE@@TWO_TRUE@am_targ_OBJECTS = two.$(OBJEXT)
208 | @ONE_FALSE@@THREE_FALSE@@TWO_FALSE@am_targ_OBJECTS =
209 | @ONE_FALSE@@THREE_TRUE@@TWO_TRUE@am_targ_OBJECTS = two.$(OBJEXT) \
210 | @ONE_FALSE@@THREE_TRUE@@TWO_TRUE@ three.$(OBJEXT)
211 | @ONE_FALSE@@THREE_TRUE@@TWO_FALSE@am_targ_OBJECTS = three.$(OBJEXT)
212 | @ONE_TRUE@@THREE_FALSE@@TWO_TRUE@am_targ_OBJECTS = one.$(OBJEXT) \
213 | @ONE_TRUE@@THREE_FALSE@@TWO_TRUE@ two.$(OBJEXT)
214 | @ONE_TRUE@@THREE_FALSE@@TWO_FALSE@am_targ_OBJECTS = one.$(OBJEXT)
215 | @ONE_TRUE@@THREE_TRUE@@TWO_TRUE@am_targ_OBJECTS = one.$(OBJEXT) \
216 | @ONE_TRUE@@THREE_TRUE@@TWO_TRUE@ two.$(OBJEXT) three.$(OBJEXT)
217 | @ONE_TRUE@@THREE_TRUE@@TWO_FALSE@am_targ_OBJECTS = one.$(OBJEXT) \
218 | @ONE_TRUE@@THREE_TRUE@@TWO_FALSE@ three.$(OBJEXT)
220 why don't we just output
222 | @ONE_TRUE@am_SONE_OBJECTS = one.$(OBJEXT)
223 | @ONE_FALSE@am_SONE_OBJECTS =
225 | @TWO_TRUE@am_STWO_OBJECTS = two.$(OBJEXT)
226 | @TWO_FALSE@am_STWO_OBJECTS =
228 | @THREE_TRUE@am_STHREE_OBJECTS = three.$(OBJEXT)
229 | @THREE_FALSE@am_STHREE_OBJECTS =
231 | am_targ_OBJECTS = $(am_SONE_OBJECTS) $(am_STWO_OBJECTS) $(am_STHREE_OBJECTS)
233 which means also, why do we look for the closure of PROGRAMS, instead
234 of just adding $(EXEEXT) to all its components and sub components
235 (i.e., inside sub vars such as $(SONE) above being a sub var of
239 Aaaaaaaaaaah! I think I know... Must be because of `+='.
241 Hm... No. Indeed we transform
250 | @BAR_TRUE@FOO = foo bar
251 | @BAR_FALSE@FOO = foo
253 but this seems good to me too?
255 | FOO = foo $(BAR_FOO)
256 | @BAR_TRUE@BAR_FOO = bar
257 | @BAR_FALSE@BAR_FOO =
264 ... this ought to work. The fix is probably complicated, but might
265 come for free when we rewrite the handling of conditionals.
267 * `distcheck' and `dist' should depend on `all'
269 * Add code to generate foo-config script like gnome, gtk
271 * `DEFS += foo' won't work.
272 That's because DEFS is defined in header-vars.am, which is read
273 after the user's Makefile.am.
274 This will be a problem for any macro defined internally
275 [ fixing this will probably fix the nasty `exeext redefines
276 foo_PROGRAMS' hack that is in there right now ]
277 [ we currently give an error when this occurs, so this is very low
280 * document user namespace for macro/target names
281 adopt some conventions and use uniformly
282 [ this is a good thing for the rewrite ]
284 * make distcheck uses directories like `=build'.
285 Some (very rare) POSIX systems don't support `=' in filenames.
286 If this ever becomes a problem, fix it
288 * distclean must remove config.status
289 can't this cause problems for maintainer-clean?
290 shouldn't maintainer-clean print the message before running
291 any part of the make? (just to slow things down long enough
292 for the user to stop it)
293 (maybe doesn't matter since people who even know about
294 maintainer-clean already have a clue)
296 * reintroduce AM_FUNC_FNMATCH which sets LIBOBJS
297 Then have automake know about fnmatch.h.
298 [ probably should wait for autoconf to get right functionality ]
300 * "make diff" capability
301 look at gcc's Makefile.in to see what to do
302 or look at maint program
304 * in --cygnus, clean-info not generated at top level
306 * what if an element of a scanned variable looks like
308 or some other arbitrary thing?
309 right now we try to cope, but not very well
310 [ this is only of theoretical interest for now ]
312 * make sure every variable that is used is also defined
313 [ we don't really look at variable uses in detail.
316 * make sure `missing' defines are generated
318 * missing should handle install -d and rmdir -p (for uninstall)
320 * notice when a .c file is a target somewhere, and auto-add it to
323 * NORMAL_INSTALL / NORMAL_UNINSTALL -vs- recursive rules
324 [ requires changes to the standard ]
326 * copyrights on m4 files, aclocal output
328 * should not put texiname_TEXINFOS into distribution
329 should rename this macro anyway, to foo_texi_DEPENDENCIES
331 * *all* installed scripts should support --version, --help
333 * For now I guess I'll just have automake give an error if it encounters
334 non-C source in a libtool library specification.
336 * must split $obj into two parts: one for libtool and one for
337 deansification. Otherwise .S files will be deansified!
339 * ansi2knr must currently appear in a directory that has some source
341 * if program has the same name as a target, do something sensible:
342 - if the target is internal, rename it
343 - if the target is mandated (eg, "info"), tell the user
344 consider auto-modifying the program name to work around this
346 * should separate actual options from strictness levels
347 strictness should only cover requirements
348 You should be able to pick and choose options
350 * rewrite in guile (RMS request)
351 at the same time, consider adding a GUI
352 could use the same parsing code for the GUI and the standalone version
353 that means figuring out a better representation of internal state
354 [ that's easy -- anything is better than what we have now ]
356 having just one Makefile for a project would give a big speed increase
357 for a project with many directories, eg glibc. ideally (?) you'd
358 still be able to have a Makefile.am in each directory somehow; this
359 might make editing conceptually easier.
361 * finish up TAGS work
363 * only remove libtool at top level?
365 * clean up source directory by moving stuff into subdirs
367 * consider adding pkglibexecdir, maybe others?
368 requests for pkg-dirs with version included
370 Avoid loops when installing; instead unroll them in automake
373 * completely handle multi-":" mode for AC_CONFIG_HEADER
374 * Scan multiple input files when Makefile is generated?
375 This would provide flexibility for large projects; subsumes
376 the "Makefile.tmpl" idea
378 [ can't do this. must explain why in manual.
379 basically, solving all the problems is too hard
380 like: how to remove redundancies between generated .in files
381 instead should implement `include' directive for Makefile.am ]
382 * for multi-":" mode and AC_OUTPUT, it might be good to pick the
383 first input file that has a corresponding .am file.
385 Some long-term projects:
386 * if $(FOO) is used somewhere, ensure FOO is defined, either by
387 user or by automake if possible
389 [ include, += support ]
390 * even better would be allowing targets in different included
391 fragments to be merged. e.g., `install-local'.
393 consider putting all check-* targets onto @check?
394 To support --help/--version checking?
396 take diff-n-query code from libit
399 Per> 1) Being able to build a set of non-source programs
400 Per> from source programs, without necessarily linking them together.
401 Per> I.e. one should be able to say something like:
402 Per> dummy_SOURCES=foo.c bar.c
403 Per> and automake should realize that it needs to build foo.o and bar.o.
404 Per> 2) Being intelligent about new kinds of suffixes.
406 Per> SUFFIXES = .class .java
407 Per> and a suffix rule of the form:
409 Per> then it should be able to realize it can build .class files from
410 Per> .java files, and thus be able to generate a list of
411 Per> .class files from a list of .java source files.
413 !! Must fix require_file stuff. It is really gross, and I don't
414 understand it any more.
416 Jim's idea: should look for @setfilename and warn if filenames too long
419 from joerg-martin schwarz:
420 -- If Makefile.am contains $(CC), $(COMPILE), $(YLWRAP), ....
421 in an explicitly written rule, you should emit the corresponding
422 Makefile variables automatically.
424 Configuring in the large:
425 * allow hierarchy of dirs to share one aclocal.m4
428 consider printing full file name of Makefile.am or configure.in when
429 giving error. This would help for very large trees with many
432 From the GNU Standards. These things could be checked, and probably
434 * Make sure that the directory into which the distribution unpacks (as
435 well as any subdirectories) are all world-writable (octal mode 777).
436 * Make sure that no file name in the distribution is more than 14
438 * Don't include any symbolic links in the distribution itself.
440 * Make sure that all the files in the distribution are world-readable.
441 ** also, check --help output and --version output. Idea from François
442 * standards no longer prohibit ANSI C. What does this imply
443 for the de-ansi-fication feature? [ must keep it -- some users rely on it ]
445 should be able to determine what is built by looking at rules (and
446 configure.in). Then built man pages (eg) could automatically be
447 omitted from the distribution.
449 Henrik Frystyk Nielsen says:
450 Henrik> 4) Flags like --include-deps are lost when you make changes to
451 Henrik> Makefile.am files and automake is run automatically. It would
452 Henrik> be nice to keep these flags as I now have to redo everything
454 ... what about other options here too?
456 Think about: maybe "make check" should just bomb if error occurs?
457 Then user must use "make -k check". This is probably more natural.
459 Consider: "cvs" option adds some cvs-specific rules?
461 Right now, targets generated internally (eg "install") are not
462 overridable by user code. This should probably be possible, even
463 though it isn't very important. This could be done by generating all
464 internal rules via a function call instead of just appending to
466 [ this will be harder to implement when scanning a rule like all-recursive
469 * Should be a way to have "nobuild_PROGRAMS" which aren't even built,
470 but which could be by running the magic make command.
473 * Must rewrite am_install_var. Should break into multiple functions.
474 This will allow the callers to be a little smarter.
475 * Rewrite clean targets.
476 * Must rewrite error handling code. Right now it is a real mess
477 Should fix up require_file junk at the same time
479 djm wants ``LINKS'' variable; list of things to link together after
480 install. In BSD environment, use:
481 LINKS = from1 to1 from2 to2 ...
483 Need way to say there are no suffixes in a Makefile (Franc,ois'
484 "override" idea suffices here)
486 Check to make sure various scripts are executable (IE when looking for
489 Handle dist-zoo. Generally add more DOS support. Maybe run "doschk"
490 (why isn't this merged with "pathchk"?) when doing a dist. Do
491 whatever else François says here...
493 Add support for html via an option. Use texi2html. Use
494 "html_TEXINFOS", and htmldir = .../html. Include html files in
495 distribution. Also allow "html_DATA", for raw .html files.
496 [ when will texinfo directly support html? ]
498 uninstall and pkg-dirs should rm -rf the dir.
500 a potential bug: configure puts "blah.o" into LIBOBJS, thus implying
501 these files can't be de-ansified. Not a problem?
502 [ fix by using ansi2knr wrapper program ]
504 In general most .am files should be merged into automake. For
505 instance all the "clean" targets could be merged by keeping lists of
506 things to be removed. This would be a lot nicer looking. Note that
507 the install targets probably should not be merged; it is sometimes
508 useful to only install a small part.
510 * Clean up the output:
511 ** Order rules sensibly
512 ** Ensure every line has a purpose. Omit unused stuff
513 ** Eliminate extraneous rules when possible (eg 'install-am' stuff)
514 ** Make sure vertical spacing is correct
515 Omit program transform vars from header if no program installed. This
516 is currently pretty hard to do. (But with beautification code it
517 would probably be easy)
520 ** It would be nice to automatically support using bison's better features
521 to rename the output files. This requires autoconf support
522 ** Consider supporting syntax from autoconf "derived:source", eg:
524 for yacc and lex source
525 ** what if you use flex and the option to avoid -lfl?
528 * Multi-language support:
529 ** should have mapping of file extensions to languages
530 ** should automatically handle the linking issue (special-case C++)
531 ** must get compile rules for various languages; FORTRAN probably
532 most important unimplemented language
533 This should be integrated in some way with Per's idea.
534 Eg .f.o rules should be recognized & auto-handled in _SOURCES
535 That way any random language can be treated with C/C++ on a first-class
538 It might be cool to generate .texi dependencies by grepping for
539 @include. (If done, it should be done the same way C dependencies are
542 It would be good to check some parts of GNU standards. Already check
543 for install-sh and mkinstalldirs. What else is required to be in
544 package by GNU standards or by automake?
545 Some things for --strictness=gnits:
546 * "cd $(foo); something" is an error in a rule. Should be:
547 "cd $(foo) && something"
548 * Look for 'ln -s' and warn about using $(LN) and AC_PROG_LN_S
549 * Look for $(LN) and require AC_PROG_LN_S
551 Auto-distribute "ChangeLog.[0-9]+"? "ChangeLog.[a-z]+"?
553 Internationalize. [ gettext doesn't have the necessary machinery yet ]
554 am_error should use printf-style arguments (for eventual gettext scheme)
556 François says the ordering of files in a distribution should be as follows:
560 I agree, but I don't see how to implement this yet.
561 It might be easier if "derived files" is limited to those that
562 Automake itself knows about, eg output of yacc.
564 Check all source files to make sure that FSF address is up-to-date.
565 --gnits or --gnu only.
567 Merge each -vars.am file with corresponding ".am" file. Can do this
568 because of changes to &file_contents.
570 Should libexec programs have the name transform done on them?
572 Order the output rules sensibly, so FOO_SOURCES and FOO_OBJECTS are
573 together and rules are in the usual order.
575 Make the output minimal: only output definitions for variables that
579 David> To avoid comments like the one about subdirs getting buried in
580 David> the middle of a Makefile.in, how about pushing comments that
581 David> start with ### to the top of the Makefile.in (in order)? Sort
582 David> of like how Autoconf uses diversions to force initialization
583 David> code to the top of configure.
586 Karl> 2) Your Makefile variable names are generally uppercase, but GNU
587 Karl> generally uses lowercase. Not that it matters :-).
589 ================================================================
593 probably should put each group of m4 files into a subdir owned by the
594 containing application.
596 ================================================================
602 how to use the generated makefiles
605 - NORMAL_INSTALL junk
607 what goes in AC_CONFIG_AUX_DIR
609 multi-":" mode in AC_OUTPUT -- automake only looks at the first file
610 also a note on how a .am file is found in this case
612 rationale for avoiding
613 make CFLAGS="$CFLAGS" ...
616 a package that installs its own aclocal macros
618 write example of using automake with dejagnu
619 follow calc example in dejagnu docs
621 document which variables are actually scanned and which are not.
623 Document customary ordering of Makefile.am. From François.
625 Should include extended version of diagram from Autoconf (suggested by
628 Make a definition of the term "source"
630 document how to use Automake with CVS. Idea from Mark Galassi. Also
631 include Greg Woods' more sophisticated "cvs-dist" target.
633 document rebuilding configure. CONFIGURE_DEPENDENCIES
634 CONFIG_STATUS_DEPENDENCIES
636 -- must document all variables that are supposed
637 to be public knowledge
639 must document the targets required for integration with
640 non-automake-using subdirs
642 document the "make SHELL='/bin/sh -x'" trick for debugging
644 section on relationship to GNU make. include notes on parallel makes
648 move discussion of cygwin32, etags, mkid under other gnu tools
652 ================================================================
654 Things to do for gcc:
656 Regularize dependency generation. Add new flags:
658 -MH Generate a dummy dependency for each header file mentioned.
662 Set name of output file
664 Then automake can use -MD -MH -MT 'foo.o foo.lo' -MF .deps/...
666 ================================================================
670 * Should support standalone library along with subdir library in same
671 Makefile.am. Maybe: turn off "standalone" mode if library's Makefile.am
672 is not only one specd? [ add an option for this ]
674 ================================================================
678 Would it be useful to integrate in some way with the Debian package
679 building utility? Must check. maybe it would be possible to deal
680 with all the different package utilities somehow. Lately I've been
681 hearing good things about the RedHat packaging utilities. Why are
682 there so many of these? Are they fun to write or something?
683 The RedHat package utility is called RPM; see
684 ftp://ftp.redhat.com/pub/code/rpm
685 It actually has problems, like no configure script and no documentation.
687 For Cygnus it would probably be good to be able to handle the native
688 package utility on each platform. There are probably 3 or 4 of these
689 (sysv, solaris?, aix?)
691 tcl/unix/Makefile.in has some code to generate a Solaris package.
693 Automake probably can't do all of this on its own. A new tool might
696 I have some notes from a Debian developer on how the integration
699 ================================================================
701 A tool to guess what the local Makefile.am should look like:
702 (see Gord's Maint program!)
704 * Probably integrate with autoscan
705 * Use various simple rules to determine what to do:
706 * get name of top directory, sans version info
707 * search for .c files with 'main' in them
708 * if in main.c, use directory name for program
709 * if in more than one, generate multiple programs
710 * if not found, generate a library named after directory
711 * order subdir searches correctly: lib first, src last
712 * assume 'testsuite' dir means we are using dejagnu
713 * maybe be smart about reading existing Makefile.am, so tool
714 can be run for incremental changes? You could imagine:
717 autoproject --incremental
719 ================================================================
721 Stuff NOT to do, and why:
723 consider auto-including any file that matches "*.in".
724 [ no: po/Makefile.in shouldn't be included ]
726 must look at mkid to see how it works (for subdir usage)
727 [ right now, it doesn't. i don't see a simple fix right now ]
729 if configure.in not found, move up a directory and try again? This
730 could eliminate a common source of problems.
731 [ this is just a bad idea ]
733 * scripts are installed in $exec_prefix/bin, not $prefix/bin
735 [ the consensus on Gnits is that this isn't required.
736 doubters can work around it anyway ]
738 Scan source directories and warn about missing files, eg .c/.h files
739 that aren't mentioned?
740 [ distcheck makes this less useful ]
743 - how to install file with a space in its name?
744 [ don't bother with this -- make is just too losing ]
748 Copyright 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001 Free Software
751 This file is part of GNU Automake.
753 GNU Automake is free software; you can redistribute it and/or modify
754 it under the terms of the GNU General Public License as published by
755 the Free Software Foundation; either version 2, or (at your option)
758 GNU Automake is distributed in the hope that it will be useful,
759 but WITHOUT ANY WARRANTY; without even the implied warranty of
760 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
761 GNU General Public License for more details.
763 You should have received a copy of the GNU General Public License
764 along with autoconf; see the file COPYING. If not, write to
765 the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
766 Boston, MA 02111-1307, USA.