1 * something like this fails:
5 lib_LTLIBRARIES = foo.la
7 this should give an error.
9 * dependency tracking doesn't work well when a file is removed
10 the new code to track header dependencies exacerbates this
12 ANSWER: for each dependency foo.h, make a dummy target `foo.h:'
13 in the .P file. Then the right thing ought to happen
15 * Every program foo has FOOFLAGS right now.
16 It should also have AM_FOOFLAGS, which can be set in Makefile.am.
18 * when cleaning, should recurse depth first
21 - how to install file with a space in its name?
23 * "make diff" capability
24 look at gcc's Makefile.in to see what to do
25 or look at maint program
27 * Karl wants to be able to set LIBS and LDFLAGS at build time, like CFLAGS
28 maybe we need something more general?
30 * add $(srcdir)/ before some dependencies?
32 * mention PACKAGE/VERSION -vs- autoheader in manual
34 * define LINK if a program is mentioned, even if no C sources appear
36 * BUILT_SOURCES should not be distributed, even when they appear in
37 another _SOURCES line. [? or maybe just leave this up to the
38 to-be-defined generic distribution method ]
39 must completely revisit the entire BUILT_SOURCES idea
41 * in --cygnus, clean-info not generated at top level
43 * what if an element of a scanned variable looks like
45 or some other arbitrary thing?
46 right now we try to cope, but not very well
48 * if `interlock' exists, that should be an error (?)
49 should also warn about using new ylwrap and not old one
50 only do this when looking for ylwrap
52 ** make sure every variable that is used is also defined
54 * make sure `missing' defines are generated
55 * if no AM_INIT_AUTOMAKE, then don't handle `missing' stuff.
57 * missing should handle install -d and rmdir -p (for uninstall)
59 * allow 'cygnus' in AUTOMAKE_OPTIONS
61 * a couple ways to be smarter:
62 - notice when a .c file is a target somewhere, and auto-add it to
64 - notice a target of the form `.x.y:' and assume it is a suffix rule
67 * NORMAL_INSTALL / NORMAL_UNINSTALL -vs- recursive rules
68 [ requires changes to the standard ]
70 * cross-compilation support:
71 programs built and used by the build process need to be
72 built for CC_FOR_BUILD
73 introduce a new variable for this
75 * if foo.y is a source, foo.h isn't auto-distributed?
77 * copyrights on m4 files, aclocal output
79 * is there a way to add a directory and then have "make" do all the
82 * put standards.texi into distribution
85 * should not put texiname_TEXINFOS into distribution
86 should rename this macro anyway, to foo_texi_DEPENDENCIES
88 * *all* installed scripts should support --version, --help
90 For now I guess I'll just have automake give an error if it encounters
91 non-C source in a libtool library specification.
93 * must split $obj into two parts: one for libtool and one for
94 deansification. Otherwise .S files will be deansified!
96 * ansi2knr must currently appear in a directory that has some source
98 * if program has the same name as a target, do something sensible:
99 - if the target is internal, rename it
100 - if the target is mandated (eg, "info"), tell the user
101 consider auto-modifying the program name to work around this
103 * should separate actual options from strictness levels
104 strictness should only cover requirements
105 You should be able to pick and choose options
107 should clean up texinfos.am; one rule is repeated 3 times, but
110 should always use perl -w
112 rewrite in guile (RMS request)
113 at the same time, consider adding a GUI
114 could use the same parsing code for the GUI and the standalone version
115 that means figuring out a better representation of internal state
116 [ that's easy -- anything is better than what we have now ]
118 having just one Makefile for a project would give a big speed increase
119 for a project with many directories, eg glibc. ideally (?) you'd
120 still be able to have a Makefile.am in each directory somehow; this
121 might make editing conceptually easier.
123 * finish up TAGS work
125 * put parser.h into distribution if "yacc -d" is used
127 * only remove libtool at top level?
129 * clean up source directory by moving stuff into subdirs
131 * consider adding pkglibexecdir, maybe others?
132 requests for pkg-dirs with version included
137 Avoid loops when installing; instead unroll them in automake
140 * completely handle multi-":" mode for AC_CONFIG_HEADER
141 * Scan multiple input files when Makefile is generated?
142 This would provide flexibility for large projects; subsumes
143 the "Makefile.tmpl" idea
145 [ can't do this. must explain why in manual.
146 basically, solving all the problems is too hard
147 like: how to remove redundancies between generated .in files
148 instead should implement `include' directive for Makefile.am ]
149 * for multi-":" mode and AC_OUTPUT, it might be good to pick the
150 first input file that has a corresponding .am file.
152 Some long-term projects:
153 * if $(FOO) is used somewhere, ensure FOO is defined, either by
154 user or by automake if possible
155 * Don't rearrange order of `include' lines relative to += assignments.
156 * Handle += assignments at all.
157 * Handle `include' lines by scanning other files, and adding
158 to Makefile.in dependency
160 consider putting all check-* targets onto @check?
161 To support --help/--version checking?
163 take diff-n-query code from libit
166 Per> 1) Being able to build a set of non-source programs
167 Per> from source programs, without necessarily linking them together.
168 Per> I.e. one should be able to say something like:
169 Per> dummy_SOURCES=foo.c bar.c
170 Per> and automake should realize that it needs to build foo.o and bar.o.
171 Per> 2) Being intelligent about new kinds of suffixes.
173 Per> SUFFIXES = .class .java
174 Per> and a suffix rule of the form:
176 Per> then it should be able to realize it can build .class files from
177 Per> .java files, and thus be able to generate a list of
178 Per> .class files from a list of .java source files.
180 !! Must fix require_file stuff. It is really gross, and I don't
181 understand it any more.
183 * error messages should print ``[info blah blah]'' command when a
184 certain part of the standards apply. saw idea in message from
185 Craig Burley. wouldn't it be really cool if compile-mode in Emacs
186 understood this convention, and you could click on such text to
187 go to the appropriate info page?
189 Jim's idea: should look for @setfilename and warn if filenames too long
192 ** many requests for a way to omit a file from the distribution.
193 Should be done like `!foo' or `~foo' in _SOURCES, etc.
194 Such files should be removed explicitly after the copy step!
195 Doing this requires rewriting macros before generating Makefile.in.
197 from joerg-martin schwarz:
198 -- If Makefile.am contains $(CC), $(COMPILE), $(YLWRAP), ....
199 in an explicitly written rule, you should emit the corresponding
200 Makefile variables automatically.
202 Configuring in the large:
203 * allow hierarchy of dirs to share one aclocal.m4
206 consider printing full file name of Makefile.am or configure.in when
207 giving error. This would help for very large trees with many
210 From the GNU Standards. These things could be checked, and probably
212 * Make sure that the directory into which the distribution unpacks (as
213 well as any subdirectories) are all world-writable (octal mode 777).
214 * Make sure that no file name in the distribution is more than 14
216 * Don't include any symbolic links in the distribution itself.
218 * Make sure that all the files in the distribution are world-readable.
219 ** also, check --help output and --version output. Idea from François
220 * standards no longer prohibit ANSI C. What does this imply
221 for the de-ansi-fication feature?
223 consider supporting "var+= stuff" syntax. rewrite to just var=... on
224 output. This is sometimes convenient when you want to write a
225 Makefile.am in more-or-less modular parts
227 should be able to determine what is built by looking at rules (and
228 configure.in). Then built man pages (eg) could automatically be
229 omitted from the distribution.
231 Idea from Joerg-Martin Schwarz: allow passing different -D flags to
232 different compiles. This can be done, but with the restriction that a
233 .c cannot appear in 2 different "objects" (programs/libraries)
234 compiled with different -D options (because -c and -o do not always
235 work together and parallel makes must work). This could be
236 implemented by noticing whenever a ".o" target with no rules is being
237 emitted, and adding the appropriate compilation rule as appropriate.
238 This should work with targets from Makefile.am as well as from .P
239 files, which means rewriting so that the Makefile.am contents aren't
240 copied into the output immediately.
241 [ this could be probably done more directly by examining the sources
242 as we scan Makefile.am ]
244 Henrik Frystyk Nielsen says:
245 Henrik> 4) Flags like --include-deps are lost when you make changes to
246 Henrik> Makefile.am files and automake is run automatically. It would
247 Henrik> be nice to keep these flags as I now have to redo everything
249 ... what about other options here too?
251 Think about: maybe "make check" should just bomb if error occurs?
252 Then user must use "make -k check". This is probably more natural.
254 Consider: "cvs" option adds some cvs-specific rules?
256 Right now, targets generated internally (eg "install") are not
257 overridable by user code. This should probably be possible, even
258 though it isn't very important. This could be done by generating all
259 internal rules via a function call instead of just appending to
261 [ this will be harder to implement when scanning a rule like all-recursive
264 * Should be a way to have "nobuild_PROGRAMS" which aren't even built,
265 but which could be by running the magic make command.
268 * Must rewrite am_install_var. Should break into multiple functions.
269 This will allow the callers to be a little smarter.
270 * Rewrite clean targets.
271 * Must rewrite error handling code. Right now it is a real mess
272 Should fix up require_file junk at the same time
274 djm wants ``LINKS'' variable; list of things to link together after
275 install. In BSD environment, use:
276 LINKS = from1 to1 from2 to2 ...
278 Need way to say there are no suffixes in a Makefile (Franc,ois'
279 "override" idea suffices here)
281 Check to make sure various scripts are executable (IE when looking for
284 Use recode in dist target when MAINT_CHARSET specified. Read caveats
285 in automake.in before doing this. Note the same problem used to apply
286 to the no-dependencies option; maybe it still should? Note also that
287 each Makefile.am must be rewritten at "make dist" time if
288 MAINT_CHARSET and DIST_CHARSET are not identical. NOTE: gettext must
289 arrange for all .po files not to be recoded. In the long term this
290 might be a problem (consider when some systems use Unicode but the
292 MAINT_CHARSET *must* be local to each Makefile.am, to enable
293 merged distributions.
294 DIST_CHARSET must be passed down to subdir makes during a "make dist"
296 Handle dist-zoo. Generally add more DOS support. Maybe run "doschk"
297 (why isn't this merged with "pathchk"?) when doing a dist. Do
298 whatever else François says here...
300 Add support for html via an option. Use texi2html. Use
301 "html_TEXINFOS", and htmldir = .../html. Include html files in
302 distribution. Also allow "html_DATA", for raw .html files.
303 [ when will texinfo directly support html? ]
305 uninstall and pkg-dirs should rm -rf the dir.
307 a potential bug: configure puts "blah.o" into LIBOBJS, thus implying
308 these files can't be de-ansified. Not a problem?
309 [ fix by using ansi2knr wrapper program ]
311 In general most .am files should be merged into automake. For
312 instance all the "clean" targets could be merged by keeping lists of
313 things to be removed. This would be a lot nicer looking. Note that
314 the install targets probably should not be merged; it is sometimes
315 useful to only install a small part.
318 * Order rules sensibly
319 * Ensure every line has a purpose. Omit unused stuff
320 * Eliminate extraneous rules when possible (eg 'install-am' stuff)
321 * Make sure vertical spacing is correct
322 Omit program transform vars from header if no program installed. This
323 is currently pretty hard to do. (But with beautification code it
324 would probably be easy)
327 * It would be nice to automatically support using bison's better features
328 to rename the output files. This requires autoconf support
329 * Consider supporting syntax from autoconf "derived:source", eg:
331 for yacc and lex source
332 * what if you use flex and the option to avoid -lfl?
335 Multi-language support:
336 * should have mapping of file extensions to languages
337 * should automatically handle the linking issue (special-case C++)
338 * must get compile rules for various languages; FORTRAN probably
339 most important unimplemented language
340 This should be integrated in some way with Per's idea.
341 Eg .f.o rules should be recognized & auto-handled in _SOURCES
342 That way any random language can be treated with C/C++ on a first-class
345 It might be cool to generate .texi dependencies by grepping for
346 @include. (If done, it should be done the same way C dependencies are
349 It would be good to check some parts of GNU standards. Already check
350 for install-sh and mkinstalldirs. What else is required to be in
351 package by GNU standards or by automake?
352 Some things for --strictness=gnits:
353 * "cd $(foo); something" is an error in a rule. Should be:
354 "cd $(foo) && something"
355 * Look for 'ln -s' and warn about using $(LN) and AC_PROG_LN_S
356 * Look for $(LN) and require AC_PROG_LN_S
358 Auto-distribute "ChangeLog.[0-9]+"? "ChangeLog.[a-z]+"?
360 Internationalize. [ gettext doesn't have the necessary machinery yet ]
361 am_error should use printf-style arguments (for eventual gettext scheme)
363 François says the ordering of files in a distribution should be as follows:
367 I agree, but I don't see how to implement this yet.
368 It might be easier if "derived files" is limited to those that
369 Automake itself knows about, eg output of yacc.
371 Check all source files to make sure that FSF address is up-to-date.
372 --gnits or --gnu only.
374 Merge each -vars.am file with corresponding ".am" file. Can do this
375 because of changes to &file_contents.
377 Should libexec programs have the name transform done on them?
379 Order the output rules sensibly, so FOO_SOURCES and FOO_OBJECTS are
380 together and rules are in the usual order.
382 Make the output minimal: only output definitions for variables that
386 David> To avoid comments like the one about subdirs getting buried in
387 David> the middle of a Makefile.in, how about pushing comments that
388 David> start with ### to the top of the Makefile.in (in order)? Sort
389 David> of like how Autoconf uses diversions to force initialization
390 David> code to the top of configure.
393 Karl> 2) Your Makefile variable names are generally uppercase, but GNU
394 Karl> generally uses lowercase. Not that it matters :-).
396 ================================================================
400 probably should put each group of m4 files into a subdir owned by the
401 containing application.
403 ================================================================
409 how to use the generated makefiles
412 - NORMAL_INSTALL junk
414 what goes in AC_CONFIG_AUX_DIR
416 multi-":" mode in AC_OUTPUT -- automake only looks at the first file
417 also a note on how a .am file is found in this case
419 rationale for avoiding
420 make CFLAGS="$CFLAGS" ...
423 a package that installs its own aclocal macros
425 write example of using automake with dejagnu
426 follow calc example in dejagnu docs
428 document which variables are actually scanned and which are not.
430 Document customary ordering of Makefile.am. From François.
432 Should include extended version of diagram from Autoconf (suggested by
435 Make a definition of the term "source"
437 document how to use Automake with CVS. Idea from Mark Galassi. Also
438 include Greg Woods' more sophisticated "cvs-dist" target.
440 document rebuilding configure. CONFIGURE_DEPENDENCIES
442 -- must document all variables that are supposed
443 to be public knowledge
445 must document the targets required for integration with
446 non-automake-using subdirs
448 document the "make SHELL='/bin/sh -x'" trick for debugging
450 section on relationship to GNU make. include notes on parallel makes
454 move discussion of cygwin32, etags, mkid under other gnu tools
456 ================================================================
458 Things to do for autoconf:
460 * patch autoreconf to run automake and aclocal. I've done this but it is
461 not really available. It can't be made available until automake
462 is officially released
464 ================================================================
468 * Should support standalone library along with subdir library in same
469 Makefile.am. Maybe: turn off "standalone" mode if library's Makefile.am
470 is not only one specd? [ add an option for this ]
472 ================================================================
476 Would it be useful to integrate in some way with the Debian package
477 building utility? Must check. maybe it would be possible to deal
478 with all the different package utilities somehow. Lately I've been
479 hearing good things about the RedHat packaging utilities. Why are
480 there so many of these? Are they fun to write or something?
481 The RedHat package utility is called RPM; see
482 ftp://ftp.redhat.com/pub/code/rpm
483 It actually has problems, like no configure script and no documentation.
485 For Cygnus it would probably be good to be able to handle the native
486 package utility on each platform. There are probably 3 or 4 of these
487 (sysv, solaris?, aix?)
489 tcl/unix/Makefile.in has some code to generate a Solaris package.
491 Automake probably can't do all of this on its own. A new tool might
494 I have some notes from a Debian developer on how the integration
497 ================================================================
499 A tool to guess what the local Makefile.am should look like:
500 (see Gord's Maint program!)
502 * Probably integrate with autoscan
503 * Use various simple rules to determine what to do:
504 * get name of top directory, sans version info
505 * search for .c files with 'main' in them
506 * if in main.c, use directory name for program
507 * if in more than one, generate multiple programs
508 * if not found, generate a library named after directory
509 * order subdir searches correctly: lib first, src last
510 * assume 'testsuite' dir means we are using dejagnu
511 * maybe be smart about reading existing Makefile.am, so tool
512 can be run for incremental changes? You could imagine:
515 autoproject --incremental
517 ================================================================
519 Stuff NOT to do, and why:
521 consider auto-including any file that matches "*.in".
522 [ no: po/Makefile.in shouldn't be included ]
524 must look at mkid to see how it works (for subdir usage)
525 [ right now, it doesn't. i don't see a simple fix right now ]
527 if configure.in not found, move up a directory and try again? This
528 could eliminate a common source of problems.
529 [ this is just a bad idea ]
531 * scripts are installed in $exec_prefix/bin, not $prefix/bin
533 [ the consensus on Gnits is that this isn't required.
534 doubters can work around it anyway ]
536 * make the auto-dep code crash if GNU make not in use?
537 (doesn't it already?)
539 Looked at a program called 'ezmake', which seems to do something
540 similar. The only idea there that is possibly worth stealing is using
541 globs in definitions. Also has negations. Eg in a directory with
542 files a.c, b.c and c.c, the line:
543 foo_SOURCES = *.c ~c.c
544 would be equivalent to:
545 foo_SOURCES = a.c b.c
546 Is this worth implementing?
547 [ No... it is more reliable to spell everything out. ]
549 Scan source directories and warn about missing files, eg .c/.h files
550 that aren't mentioned?
551 [ distcheck makes this less useful ]