1 we can't seem to AC_SUBST(pkgdatadir)
2 the version from header-vars overrides
5 check should depend on all
8 the new YFLAGS code doesn't correctly handle srcdir
10 allow foo_NAME to rename an object (library or program)
13 remove _LTLIBRARIES and just use _LIBRARIES
14 then use this for zip/jar as well
16 consider `sub am_unique_error', which only prints a given error
17 message once. Use eg in lang_c_rewrite for ansi2knr error
20 investigate problems with conditionally defined libraries
22 add an error if the user makefile.am violates our
25 we need a document describing automake from the end user's point of view
26 eg describe INSTALL_HEADER there, among other things
32 > DISTCLEAN -test -z "$(DISTCLEANFILES)" || rm -f $(DISTCLEANFILES)
34 > maintainer-clean-generic:
35 > +## FIXME: shouldn't we really print these messages before running
36 > +## the dependencies?
37 > + @echo "This command is intended for maintainers to use"
38 > + @echo "it deletes files that may require special tools to rebuild."
42 > I'd like to eventually fix the FIXME comment by having
43 > maintainer-clean look like:
49 > We're left with the question of whether we should repeat them in every
54 > Hmm... Interesting. It must have been a side effect of the enabling
55 > of forced `relink' on GNU/Linux/x86. Anyway, on platforms that
56 > actually require relinking, this problem remains, and I see no way to
57 > overcome it other than arranging for automake to install libraries
58 > before executables, as you suggest. This shouldn't be a big problem,
61 > A bigger problem could show up if two libraries in the same directory,
62 > one dependent on the other, are installed concurrently. If relinking
63 > is needed for the dependent library, we have a problem. It appears to
64 > me that user will have to live without `make -j install', in this
68 > Here's an Automake patch and changelog entry allow make -j install on
69 > such degenerate systems (and Linux with buggy libtool <g>)
71 > If you install to locations other that bin_ and lib_ then a larger fix
72 > is necessary, but this should fix the 90% case.
74 * think about how per-object flags should work. in particular:
75 * how should they be specified?
76 using the object name is confusing when .lo/.obj in use
77 however, the object name provides a nice interaction with
79 * how should they interact with per-executable flags?
80 [ this is probably a feature in search of a problem ]
82 * cross-compilation support:
83 programs built and used by the build process need to be
84 built for CC_FOR_BUILD
85 introduce a new prefxi for this, e.g. `build_PROGRAMS'
86 [ we can do this in an automatic way I think.
87 unfortunately it isn't that useful until autoconf has support
88 for this sort of thing as well ]
90 * distcheck should make sure that each file that uses _() is
93 # Verify that all source files using _() are listed in po/POTFILES.in.
95 grep -E -v '^(#|$$)' po/POTFILES.in | sort > $@-1
96 grep -E -l '\b_\(' lib/*.c src/*.c | sort > $@-2
100 * one performance enhancement would be to have autoconf write
101 a single file containing all the macro assignments.
102 then read this file via `include'
103 unfortunately this can't be done because of conditionals
104 -- but it could be made to work if we also changed:
105 * automake to rewrite @FOO@ to $(FOO), and
106 * the implementation of conditionals to rely on some new
109 * support prog_LIBS as override for LIBS
111 * Test subdir-objects option with yacc, lex, ansi2knr
112 Our locking scheme won't prevent a parallel make from losing
113 if there are two `bar.o' files and the timing is just right
114 This only happens with parallel make and no-`-c -o' compiler,
115 so it probably isn't very important
116 `-c -o' when doing libtool
117 try to find a losing compiler and see if it really works.
118 (actually: hack config.cache and do it)
121 ** We're using `$<' in explicit rules when using per-exe flags
122 ** per-exe flags don't work for CPPFLAGS/YFLAGS/LFLAGS. Fix.
123 ** LIBOBJS shouldn't be used when there are per-exe flags (?)
125 * Test nodist_SOURCES with lex, yacc, etc.
127 * Support subdir-objects with fortran
129 * Allow creation of Java .zip/.jar files in natural way
130 If you are building a compiled Java library, then the .zip/.jar
131 ought to be made automatically.
133 * Run automake before libtool. It will report an error but
134 still won't put the file into the disty. This is wrong.
135 From Mark H Wilkinson <mhw@kremvax.demon.co.uk>
137 * in gnu/gnits mode, give error if Makefile.am overrides a user
138 variable like CFLAGS.
139 [ this is low priority because the package author can always
140 circumvent our check by redefining in configure.in
141 plus it is probably better to encourage good behavior than to
144 * examine possibility of using any character in a macro name
145 and rewriting names automatically. this means we must rewrite
146 all references as well.
147 [ this is a 2.0-style feature ]
149 * AM_CONFIG_HEADER might generate the wrong stamp file names
150 when given multiple headers. Write a test.
152 * Currently don't correctly handle multiple inputs to a config header.
153 [ this should no matter in the future as acconfig.h and so on are
154 obsoleted by the AH series of macros.]
156 * `distcheck' and `dist' should depend on `all'
158 * Add code to generate foo-config script like gnome, gtk
160 * document user namespace for macro/target names
161 adopt some conventions and use uniformly
162 [ this is a good thing for the rewrite ]
164 * make distcheck uses directories like `=build'.
165 Some (very rare) POSIX systems don't support `=' in filenames.
166 If this ever becomes a problem, fix it
168 * distclean must remove config.status
169 can't this cause problems for maintainer-clean?
170 shouldn't maintainer-clean print the message before running
171 any part of the make? (just to slow things down long enough
172 for the user to stop it)
173 (maybe doesn't matter since people who even know about
174 maintainer-clean already have a clue)
176 * reintroduce AM_FUNC_FNMATCH which sets LIBOBJS
177 Then have automake know about fnmatch.h.
178 [ probably should wait for autoconf to get right functionality ]
180 * "make diff" capability
181 look at gcc's Makefile.in to see what to do
182 or look at maint program
184 * in --cygnus, clean-info not generated at top level
186 * what if an element of a scanned variable looks like
188 or some other arbitrary thing?
189 right now we try to cope, but not very well
190 [ this is only of theoretical interest for now ]
192 * make sure every variable that is used is also defined
193 [ we don't really look at variable uses in detail.
196 * make sure `missing' defines are generated
198 * missing should handle install -d and rmdir -p (for uninstall)
200 * NORMAL_INSTALL / NORMAL_UNINSTALL -vs- recursive rules
201 [ requires changes to the standard ]
203 * copyrights on m4 files, aclocal output
205 * should not put texiname_TEXINFOS into distribution
206 should rename this macro anyway, to foo_texi_DEPENDENCIES
208 * *all* installed scripts should support --version, --help
210 * For now I guess I'll just have automake give an error if it encounters
211 non-C source in a libtool library specification.
213 * must split $obj into two parts: one for libtool and one for
214 deansification. Otherwise .S files will be deansified!
216 * ansi2knr must currently appear in a directory that has some source
218 * if program has the same name as a target, do something sensible:
219 - if the target is internal, rename it
220 - if the target is mandated (eg, "info"), tell the user
221 consider auto-modifying the program name to work around this
223 * should separate actual options from strictness levels
224 strictness should only cover requirements
225 You should be able to pick and choose options
227 * rewrite in guile (RMS request)
228 at the same time, consider adding a GUI
229 could use the same parsing code for the GUI and the standalone version
230 that means figuring out a better representation of internal state
231 [ that's easy -- anything is better than what we have now ]
233 having just one Makefile for a project would give a big speed increase
234 for a project with many directories, eg glibc. ideally (?) you'd
235 still be able to have a Makefile.am in each directory somehow; this
236 might make editing conceptually easier.
238 * finish up TAGS work
240 * only remove libtool at top level?
242 * clean up source directory by moving stuff into subdirs
244 * consider adding pkglibexecdir, maybe others?
245 requests for pkg-dirs with version included
247 Avoid loops when installing; instead unroll them in automake
248 [ Impossible when @AC_SUBST@ values are used. ]
251 * completely handle multi-":" mode for AC_CONFIG_HEADER
252 * Scan multiple input files when Makefile is generated?
253 This would provide flexibility for large projects; subsumes
254 the "Makefile.tmpl" idea
256 [ can't do this. must explain why in manual.
257 basically, solving all the problems is too hard
258 like: how to remove redundancies between generated .in files
259 instead should implement `include' directive for Makefile.am ]
260 * for multi-":" mode and AC_OUTPUT, it might be good to pick the
261 first input file that has a corresponding .am file.
263 Some long-term projects:
264 * if $(FOO) is used somewhere, ensure FOO is defined, either by
265 user or by automake if possible
267 [ include, += support ]
268 * even better would be allowing targets in different included
269 fragments to be merged. e.g., `install-local'.
271 consider putting all check-* targets onto @check?
272 To support --help/--version checking?
274 take diff-n-query code from libit
277 Per> 1) Being able to build a set of non-source programs
278 Per> from source programs, without necessarily linking them together.
279 Per> I.e. one should be able to say something like:
280 Per> dummy_SOURCES=foo.c bar.c
281 Per> and automake should realize that it needs to build foo.o and bar.o.
282 Per> 2) Being intelligent about new kinds of suffixes.
284 Per> SUFFIXES = .class .java
285 Per> and a suffix rule of the form:
287 Per> then it should be able to realize it can build .class files from
288 Per> .java files, and thus be able to generate a list of
289 Per> .class files from a list of .java source files.
290 [What Per wanted here was a way to have automate automatically follow
291 suffix rules. So for instance if you had a `.x.y:' rule, and automake
292 saw a `.x' file, it would automatically build and install the
293 corresponding `.y' file.]
295 !! Must fix require_file stuff. It is really gross, and I don't
296 understand it any more.
298 Jim's idea: should look for @setfilename and warn if filenames too long
301 from joerg-martin schwarz:
302 -- If Makefile.am contains $(CC), $(COMPILE), $(YLWRAP), ....
303 in an explicitly written rule, you should emit the corresponding
304 Makefile variables automatically.
306 Configuring in the large:
307 * allow hierarchy of dirs to share one aclocal.m4
310 consider printing full file name of Makefile.am or configure.in when
311 giving error. This would help for very large trees with many
314 From the GNU Standards. These things could be checked, and probably
316 * Make sure that the directory into which the distribution unpacks (as
317 well as any subdirectories) are all world-writable (octal mode 777).
318 * Make sure that no file name in the distribution is more than 14
320 * Don't include any symbolic links in the distribution itself.
322 * Make sure that all the files in the distribution are world-readable.
323 ** also, check --help output and --version output. Idea from François
324 * standards no longer prohibit ANSI C. What does this imply
325 for the de-ansi-fication feature? [ must keep it -- some users rely on it ]
327 should be able to determine what is built by looking at rules (and
328 configure.in). Then built man pages (eg) could automatically be
329 omitted from the distribution.
331 Consider: "cvs" option adds some cvs-specific rules?
333 Right now, targets generated internally (eg "install") are not
334 overridable by user code. This should probably be possible, even
335 though it isn't very important. This could be done by generating all
336 internal rules via a function call instead of just appending to
338 [ this will be harder to implement when scanning a rule like all-recursive
341 * Should be a way to have "nobuild_PROGRAMS" which aren't even built,
342 but which could be by running the magic make command.
343 [ How does it differ from EXTRA_PROGRAMS? ]
346 * Must rewrite am_install_var. Should break into multiple functions.
347 This will allow the callers to be a little smarter.
348 * Rewrite clean targets.
349 * Must rewrite error handling code. Right now it is a real mess
350 Should fix up require_file junk at the same time
352 djm wants ``LINKS'' variable; list of things to link together after
353 install. In BSD environment, use:
354 LINKS = from1 to1 from2 to2 ...
356 Need way to say there are no suffixes in a Makefile (Franc,ois'
357 "override" idea suffices here)
359 Check to make sure various scripts are executable (IE when looking for
362 Handle dist-zoo. Generally add more DOS support. Maybe run "doschk"
363 (why isn't this merged with "pathchk"?) when doing a dist. Do
364 whatever else François says here...
366 Add support for html via an option. Use texi2html. Use
367 "html_TEXINFOS", and htmldir = .../html. Include html files in
368 distribution. Also allow "html_DATA", for raw .html files.
369 [ when will texinfo directly support html? ]
371 uninstall and pkg-dirs should rm -rf the dir.
373 a potential bug: configure puts "blah.o" into LIBOBJS, thus implying
374 these files can't be de-ansified. Not a problem?
375 [ fix by using ansi2knr wrapper program ]
377 In general most .am files should be merged into automake. For
378 instance all the "clean" targets could be merged by keeping lists of
379 things to be removed. This would be a lot nicer looking. Note that
380 the install targets probably should not be merged; it is sometimes
381 useful to only install a small part.
383 * Clean up the output:
384 ** Order rules sensibly
385 ** Ensure every line has a purpose. Omit unused stuff
386 ** Eliminate extraneous rules when possible (eg 'install-am' stuff)
387 ** Make sure vertical spacing is correct
388 Omit program transform vars from header if no program installed. This
389 is currently pretty hard to do. (But with beautification code it
390 would probably be easy)
393 ** It would be nice to automatically support using bison's better features
394 to rename the output files. This requires autoconf support
395 ** Consider supporting syntax from autoconf "derived:source", eg:
397 for yacc and lex source
398 ** what if you use flex and the option to avoid -lfl?
401 * Multi-language support:
402 ** should have mapping of file extensions to languages
403 ** should automatically handle the linking issue (special-case C++)
404 ** must get compile rules for various languages; FORTRAN probably
405 most important unimplemented language
406 This should be integrated in some way with Per's idea.
407 Eg .f.o rules should be recognized & auto-handled in _SOURCES
408 That way any random language can be treated with C/C++ on a first-class
411 It might be cool to generate .texi dependencies by grepping for
412 @include. (If done, it should be done the same way C dependencies are
415 It would be good to check some parts of GNU standards. Already check
416 for install-sh and mkinstalldirs. What else is required to be in
417 package by GNU standards or by automake?
418 Some things for --strictness=gnits:
419 * "cd $(foo); something" is an error in a rule. Should be:
420 "cd $(foo) && something"
421 * Look for 'ln -s' and warn about using $(LN) and AC_PROG_LN_S
422 * Look for $(LN) and require AC_PROG_LN_S
424 Auto-distribute "ChangeLog.[0-9]+"? "ChangeLog.[a-z]+"?
426 Internationalize. [ gettext doesn't have the necessary machinery yet ]
427 am_error should use printf-style arguments (for eventual gettext scheme)
429 François says the ordering of files in a distribution should be as follows:
433 I agree, but I don't see how to implement this yet.
434 It might be easier if "derived files" is limited to those that
435 Automake itself knows about, eg output of yacc.
437 Check all source files to make sure that FSF address is up-to-date.
438 --gnits or --gnu only.
440 Merge each -vars.am file with corresponding ".am" file. Can do this
441 because of changes to &file_contents.
443 Should libexec programs have the name transform done on them?
445 Order the output rules sensibly, so FOO_SOURCES and FOO_OBJECTS are
446 together and rules are in the usual order.
448 Make the output minimal: only output definitions for variables that
452 David> To avoid comments like the one about subdirs getting buried in
453 David> the middle of a Makefile.in, how about pushing comments that
454 David> start with ### to the top of the Makefile.in (in order)? Sort
455 David> of like how Autoconf uses diversions to force initialization
456 David> code to the top of configure.
459 Karl> 2) Your Makefile variable names are generally uppercase, but GNU
460 Karl> generally uses lowercase. Not that it matters :-).
462 ================================================================
466 probably should put each group of m4 files into a subdir owned by the
467 containing application.
469 ================================================================
475 how to use the generated makefiles
478 - NORMAL_INSTALL junk
480 what goes in AC_CONFIG_AUX_DIR
482 multi-":" mode in AC_OUTPUT -- automake only looks at the first file
483 also a note on how a .am file is found in this case
485 rationale for avoiding
486 make CFLAGS="$CFLAGS" ...
489 a package that installs its own aclocal macros
491 write example of using automake with dejagnu
492 follow calc example in dejagnu docs
494 document which variables are actually scanned and which are not.
496 Document customary ordering of Makefile.am. From François.
498 Should include extended version of diagram from Autoconf (suggested by
501 Make a definition of the term "source"
503 document how to use Automake with CVS. Idea from Mark Galassi. Also
504 include Greg Woods' more sophisticated "cvs-dist" target.
506 document rebuilding configure. CONFIGURE_DEPENDENCIES
507 CONFIG_STATUS_DEPENDENCIES
509 -- must document all variables that are supposed
510 to be public knowledge
512 must document the targets required for integration with
513 non-automake-using subdirs
515 document the "make SHELL='/bin/sh -x'" trick for debugging
517 section on relationship to GNU make. include notes on parallel makes
521 move discussion of cygwin32, etags, mkid under other gnu tools
525 ================================================================
529 * Should support standalone library along with subdir library in same
530 Makefile.am. Maybe: turn off "standalone" mode if library's Makefile.am
531 is not only one specd? [ add an option for this ]
533 ================================================================
537 Would it be useful to integrate in some way with the Debian package
538 building utility? Must check. maybe it would be possible to deal
539 with all the different package utilities somehow. Lately I've been
540 hearing good things about the RedHat packaging utilities. Why are
541 there so many of these? Are they fun to write or something?
542 The RedHat package utility is called RPM; see
543 ftp://ftp.redhat.com/pub/code/rpm
544 It actually has problems, like no configure script and no documentation.
546 For Cygnus it would probably be good to be able to handle the native
547 package utility on each platform. There are probably 3 or 4 of these
548 (sysv, solaris?, aix?)
550 tcl/unix/Makefile.in has some code to generate a Solaris package.
552 Automake probably can't do all of this on its own. A new tool might
555 I have some notes from a Debian developer on how the integration
558 ================================================================
560 A tool to guess what the local Makefile.am should look like:
561 (see Gord's Maint program!)
563 * Probably integrate with autoscan
564 * Use various simple rules to determine what to do:
565 * get name of top directory, sans version info
566 * search for .c files with 'main' in them
567 * if in main.c, use directory name for program
568 * if in more than one, generate multiple programs
569 * if not found, generate a library named after directory
570 * order subdir searches correctly: lib first, src last
571 * assume 'testsuite' dir means we are using dejagnu
572 * maybe be smart about reading existing Makefile.am, so tool
573 can be run for incremental changes? You could imagine:
576 autoproject --incremental
578 ================================================================
580 Stuff NOT to do, and why:
582 consider auto-including any file that matches "*.in".
583 [ no: po/Makefile.in shouldn't be included ]
585 must look at mkid to see how it works (for subdir usage)
586 [ right now, it doesn't. i don't see a simple fix right now ]
588 if configure.in not found, move up a directory and try again? This
589 could eliminate a common source of problems.
590 [ this is just a bad idea ]
592 * scripts are installed in $exec_prefix/bin, not $prefix/bin
594 [ the consensus on Gnits is that this isn't required.
595 doubters can work around it anyway ]
597 Scan source directories and warn about missing files, eg .c/.h files
598 that aren't mentioned?
599 [ distcheck makes this less useful ]
602 - how to install file with a space in its name?
603 [ don't bother with this -- make is just too losing ]
605 * notice when a .c file is a target somewhere, and auto-add it to
607 [ BUILT_SOURCES are for files that need to be built before anything
608 else because of hidden dependencies (something .c files are
614 Copyright 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002 Free
615 Software Foundation, Inc.
617 This file is part of GNU Automake.
619 GNU Automake is free software; you can redistribute it and/or modify
620 it under the terms of the GNU General Public License as published by
621 the Free Software Foundation; either version 2, or (at your option)
624 GNU Automake is distributed in the hope that it will be useful,
625 but WITHOUT ANY WARRANTY; without even the implied warranty of
626 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
627 GNU General Public License for more details.
629 You should have received a copy of the GNU General Public License
630 along with autoconf; see the file COPYING. If not, write to
631 the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
632 Boston, MA 02111-1307, USA.