1 * update text in missing
4 * if `interlock' exists, that should be an error (?)
5 should also warn about using new ylwrap and not old one
6 only do this when looking for ylwrap
8 ** make sure every variable that is used is also defined
10 * make sure `missing' defines are generated
11 * if no AM_INIT_AUTOMAKE, then don't handle `missing' stuff.
14 * allow 'cygnus' in AUTOMAKE_OPTIONS
18 * NORMAL_INSTALL / NORMAL_UNINSTALL -vs- recursive rules
19 [ requires changes to the standard ]
21 * if foo.y is a source, foo.h isn't auto-distributed?
23 * dependency tracking doesn't work well when a file is removed
24 the new code to track header dependencies exacerbates this
26 it would probably be better to use "gcc -MD" and move the .d
27 file into the .deps directory. That is, create the dependencies
28 as a side effect of compilation
29 This still won't solve the file-deletion problem
30 [ also: jim makes distributions by checking out, configuring,
31 and running "make dist". This scheme would cause that to fail ]
33 * copyrights on m4 files, aclocal output
35 * is there a way to add a directory and then have "make" do all the
38 * put standards.texi into distribution
41 * should not put texiname_TEXINFOS into distribution
42 should rename this macro anyway, to foo_texi_DEPENDENCIES
44 * *all* installed scripts should support --version, --help
46 * have aclocal diagnose unrecognized AM_ macros
48 For now I guess I'll just have automake give an error if it encounters
49 non-C source in a libtool library specification.
51 * must split $obj into two parts: one for libtool and one for
52 deansification. Otherwise .S files will be deansified!
54 * if program has the same name as a target, do something sensible:
55 - if the target is internal, rename it
56 - if the target is mandated (eg, "info"), tell the user
57 consider auto-modifying the program name to work around this
59 * should separate actual options from strictness levels
60 strictness should only cover requirements
61 You should be able to pick and choose options
63 should clean up texinfos.am; one rule is repeated 3 times, but
66 should always use perl -w
68 rewrite in guile (RMS request)
69 at the same time, consider adding a GUI
70 could use the same parsing code for the GUI and the standalone version
71 that means figuring out a better representation of internal state
72 [ that's easy -- anything is better than what we have now ]
74 having just one Makefile for a project would give a big speed increase
75 for a project with many directories, eg glibc. ideally (?) you'd
76 still be able to have a Makefile.am in each directory somehow; this
77 might make editing conceptually easier.
81 * put parser.h into distribution if "yacc -d" is used
83 * only remove libtool at top level?
85 * clean up source directory by moving stuff into subdirs
87 * consider adding pkglibexecdir, maybe others?
88 requests for pkg-dirs with version included
93 Avoid loops when installing; instead unroll them in automake
96 * completely handle multi-":" mode for AC_CONFIG_HEADER
97 * Scan multiple input files when Makefile is generated?
98 This would provide flexibility for large projects; subsumes
99 the "Makefile.tmpl" idea
101 [ can't do this. must explain why in manual.
102 basically, solving all the problems is too hard
103 like: how to remove redundancies between generated .in files
104 instead should implement `include' directive for Makefile.am ]
105 * for multi-":" mode and AC_OUTPUT, it might be good to pick the
106 first input file that has a corresponding .am file.
108 Some long-term projects:
109 * if $(FOO) is used somewhere, ensure FOO is defined, either by
110 user or by automake if possible
111 * Don't rearrange order of `include' lines relative to += assignments.
112 * Handle += assignments at all.
113 * Handle `include' lines by scanning other files, and adding
114 to Makefile.in dependency
116 consider putting all check-* targets onto @check?
117 To support --help/--version checking?
119 take diff-n-query code from libit
122 Per> 1) Being able to build a set of non-source programs
123 Per> from source programs, without necessarily linking them together.
124 Per> I.e. one should be able to say something like:
125 Per> dummy_SOURCES=foo.c bar.c
126 Per> and automake should realize that it needs to build foo.o and bar.o.
127 Per> 2) Being intelligent about new kinds of suffixes.
129 Per> SUFFIXES = .class .java
130 Per> and a suffix rule of the form:
132 Per> then it should be able to realize it can build .class files from
133 Per> .java files, and thus be able to generate a list of
134 Per> .class files from a list of .java source files.
136 !! Must fix require_file stuff. It is really gross, and I don't
137 understand it any more.
139 * error messages should print ``[info blah blah]'' command when a
140 certain part of the standards apply. saw idea in message from
141 Craig Burley. wouldn't it be really cool if compile-mode in Emacs
142 understood this convention, and you could click on such text to
143 go to the appropriate info page?
145 Jim's idea: should look for @setfilename and warn if filenames too long
148 ** many requests for a way to omit a file from the distribution.
149 Should be done like `!foo' or `~foo' in _SOURCES, etc.
150 Such files should be removed explicitly after the copy step!
151 Doing this requires rewriting macros before generating Makefile.in.
153 from joerg-martin schwarz:
154 -- If Makefile.am contains $(CC), $(COMPILE), $(YLWRAP), ....
155 in an explicitly written rule, you should emit the corresponding
156 Makefile variables automatically.
158 Configuring in the large:
159 * allow hierarchy of dirs to share one aclocal.m4
162 consider printing full file name of Makefile.am or configure.in when
163 giving error. This would help for very large trees with many
166 From the GNU Standards. These things could be checked, and probably
168 * Make sure that the directory into which the distribution unpacks (as
169 well as any subdirectories) are all world-writable (octal mode 777).
170 * Make sure that no file name in the distribution is more than 14
172 * Don't include any symbolic links in the distribution itself.
174 * Make sure that all the files in the distribution are world-readable.
175 ** also, check --help output and --version output. Idea from François
176 * standards no longer prohibit ANSI C. What does this imply
177 for the de-ansi-fication feature?
179 consider supporting "var+= stuff" syntax. rewrite to just var=... on
180 output. This is sometimes convenient when you want to write a
181 Makefile.am in more-or-less modular parts
183 should be able to determine what is built by looking at rules (and
184 configure.in). Then built man pages (eg) could automatically be
185 omitted from the distribution.
187 Idea from Joerg-Martin Schwarz: allow passing different -D flags to
188 different compiles. This can be done, but with the restriction that a
189 .c cannot appear in 2 different "objects" (programs/libraries)
190 compiled with different -D options (because -c and -o do not always
191 work together and parallel makes must work). This could be
192 implemented by noticing whenever a ".o" target with no rules is being
193 emitted, and adding the appropriate compilation rule as appropriate.
194 This should work with targets from Makefile.am as well as from .P
195 files, which means rewriting so that the Makefile.am contents aren't
196 copied into the output immediately.
197 [ this could be probably done more directly by examining the sources
198 as we scan Makefile.am ]
200 Henrik Frystyk Nielsen says:
201 Henrik> 4) Flags like --include-deps are lost when you make changes to
202 Henrik> Makefile.am files and automake is run automatically. It would
203 Henrik> be nice to keep these flags as I now have to redo everything
205 ... what about other options here too?
207 Think about: maybe "make check" should just bomb if error occurs?
208 Then user must use "make -k check". This is probably more natural.
210 Consider: "cvs" option adds some cvs-specific rules?
212 Right now, targets generated internally (eg "install") are not
213 overridable by user code. This should probably be possible, even
214 though it isn't very important. This could be done by generating all
215 internal rules via a function call instead of just appending to
217 [ this will be harder to implement when scanning a rule like all-recursive
220 * Should be a way to have "nobuild_PROGRAMS" which aren't even built,
221 but which could be by running the magic make command.
224 * Must rewrite am_install_var. Should break into multiple functions.
225 This will allow the callers to be a little smarter.
226 * Rewrite clean targets.
227 * Must rewrite error handling code. Right now it is a real mess
228 Should fix up require_file junk at the same time
230 djm wants ``LINKS'' variable; list of things to link together after
231 install. In BSD environment, use:
232 LINKS = from1 to1 from2 to2 ...
234 Need way to say there are no suffixes in a Makefile (Franc,ois'
235 "override" idea suffices here)
237 Check to make sure various scripts are executable (IE when looking for
240 Use recode in dist target when MAINT_CHARSET specified. Read caveats
241 in automake.in before doing this. Note the same problem used to apply
242 to the no-dependencies option; maybe it still should? Note also that
243 each Makefile.am must be rewritten at "make dist" time if
244 MAINT_CHARSET and DIST_CHARSET are not identical. NOTE: gettext must
245 arrange for all .po files not to be recoded. In the long term this
246 might be a problem (consider when some systems use Unicode but the
248 MAINT_CHARSET *must* be local to each Makefile.am, to enable
249 merged distributions.
250 DIST_CHARSET must be passed down to subdir makes during a "make dist"
252 Handle dist-zoo. Generally add more DOS support. Maybe run "doschk"
253 (why isn't this merged with "pathchk"?) when doing a dist. Do
254 whatever else François says here...
256 Add support for html via an option. Use texi2html. Use
257 "html_TEXINFOS", and htmldir = .../html. Include html files in
258 distribution. Also allow "html_DATA", for raw .html files.
259 [ when will texinfo directly support html? ]
261 uninstall and pkg-dirs should rm -rf the dir.
263 a potential bug: configure puts "blah.o" into LIBOBJS, thus implying
264 these files can't be de-ansified. Not a problem?
265 [ fix by using ansi2knr wrapper program ]
267 In general most .am files should be merged into automake. For
268 instance all the "clean" targets could be merged by keeping lists of
269 things to be removed. This would be a lot nicer looking. Note that
270 the install targets probably should not be merged; it is sometimes
271 useful to only install a small part.
274 * Order rules sensibly
275 * Ensure every line has a purpose. Omit unused stuff
276 * Eliminate extraneous rules when possible (eg 'install-am' stuff)
277 * Make sure vertical spacing is correct
278 Omit program transform vars from header if no program installed. This
279 is currently pretty hard to do. (But with beautification code it
280 would probably be easy)
283 * It would be nice to automatically support using bison's better features
284 to rename the output files. This requires autoconf support
285 * Consider supporting syntax from autoconf "derived:source", eg:
287 for yacc and lex source
288 * what if you use flex and the option to avoid -lfl?
291 Multi-language support:
292 * should have mapping of file extensions to languages
293 * should automatically handle the linking issue (special-case C++)
294 * must get compile rules for various languages; FORTRAN probably
295 most important unimplemented language
296 This should be integrated in some way with Per's idea.
297 Eg .f.o rules should be recognized & auto-handled in _SOURCES
298 That way any random language can be treated with C/C++ on a first-class
301 It might be cool to generate .texi dependencies by grepping for
302 @include. (If done, it should be done the same way C dependencies are
305 It would be good to check some parts of GNU standards. Already check
306 for install-sh and mkinstalldirs. What else is required to be in
307 package by GNU standards or by automake?
308 Some things for --strictness=gnits:
309 * "cd $(foo); something" is an error in a rule. Should be:
310 "cd $(foo) && something"
311 * Look for 'ln -s' and warn about using $(LN) and AC_PROG_LN_S
312 * Look for $(LN) and require AC_PROG_LN_S
314 Auto-distribute "ChangeLog.[0-9]+"? "ChangeLog.[a-z]+"?
316 Internationalize. [ gettext doesn't have the necessary machinery yet ]
317 am_error should use printf-style arguments (for eventual gettext scheme)
319 François says the ordering of files in a distribution should be as follows:
323 I agree, but I don't see how to implement this yet.
324 It might be easier if "derived files" is limited to those that
325 Automake itself knows about, eg output of yacc.
327 Check all source files to make sure that FSF address is up-to-date.
328 --gnits or --gnu only.
330 Merge each -vars.am file with corresponding ".am" file. Can do this
331 because of changes to &file_contents.
333 Should libexec programs have the name transform done on them?
335 Order the output rules sensibly, so FOO_SOURCES and FOO_OBJECTS are
336 together and rules are in the usual order.
338 Make the output minimal: only output definitions for variables that
342 David> To avoid comments like the one about subdirs getting buried in
343 David> the middle of a Makefile.in, how about pushing comments that
344 David> start with ### to the top of the Makefile.in (in order)? Sort
345 David> of like how Autoconf uses diversions to force initialization
346 David> code to the top of configure.
349 Karl> 2) Your Makefile variable names are generally uppercase, but GNU
350 Karl> generally uses lowercase. Not that it matters :-).
352 ================================================================
356 probably should put each group of m4 files into a subdir owned by the
357 containing application.
359 ================================================================
365 how to use the generated makefiles
368 - NORMAL_INSTALL junk
370 what goes in AC_CONFIG_AUX_DIR
372 multi-":" mode in AC_OUTPUT -- automake only looks at the first file
373 also a note on how a .am file is found in this case
375 rationale for avoiding
376 make CFLAGS="$CFLAGS" ...
379 a package that installs its own aclocal macros
381 write example of using automake with dejagnu
382 follow calc example in dejagnu docs
384 document which variables are actually scanned and which are not.
386 Document customary ordering of Makefile.am. From François.
388 Should include extended version of diagram from Autoconf (suggested by
391 Make a definition of the term "source"
393 document how to use Automake with CVS. Idea from Mark Galassi. Also
394 include Greg Woods' more sophisticated "cvs-dist" target.
396 document rebuilding configure. CONFIGURE_DEPENDENCIES
398 -- must document all variables that are supposed
399 to be public knowledge
401 automake must be run in each directory with a configure.in
402 This is insufficiently clear
404 must document the targets required for integration with
405 non-automake-using subdirs
407 document the "make SHELL='/bin/sh -x'" trick for debugging
409 section on relationship to GNU make. include notes on parallel makes
413 ================================================================
415 Things to do for autoconf:
417 * patch autoreconf to run automake and aclocal. I've done this but it is
418 not really available. It can't be made available until automake
419 is officially released
421 ================================================================
425 * Should support standalone library along with subdir library in same
426 Makefile.am. Maybe: turn off "standalone" mode if library's Makefile.am
427 is not only one specd? [ add an option for this ]
429 ================================================================
433 Would it be useful to integrate in some way with the Debian package
434 building utility? Must check. maybe it would be possible to deal
435 with all the different package utilities somehow. Lately I've been
436 hearing good things about the RedHat packaging utilities. Why are
437 there so many of these? Are they fun to write or something?
438 The RedHat package utility is called RPM; see
439 ftp://ftp.redhat.com/pub/code/rpm
440 It actually has problems, like no configure script and no documentation.
442 For Cygnus it would probably be good to be able to handle the native
443 package utility on each platform. There are probably 3 or 4 of these
444 (sysv, solaris?, aix?)
446 tcl/unix/Makefile.in has some code to generate a Solaris package.
448 Automake probably can't do all of this on its own. A new tool might
451 I have some notes from a Debian developer on how the integration
454 ================================================================
456 A tool to guess what the local Makefile.am should look like:
457 (see Gord's Maint program!)
459 * Probably integrate with autoscan
460 * Use various simple rules to determine what to do:
461 * get name of top directory, sans version info
462 * search for .c files with 'main' in them
463 * if in main.c, use directory name for program
464 * if in more than one, generate multiple programs
465 * if not found, generate a library named after directory
466 * order subdir searches correctly: lib first, src last
467 * assume 'testsuite' dir means we are using dejagnu
468 * maybe be smart about reading existing Makefile.am, so tool
469 can be run for incremental changes? You could imagine:
472 autoproject --incremental
474 ================================================================
476 Stuff NOT to do, and why:
478 consider auto-including any file that matches "*.in".
479 [ no: po/Makefile.in shouldn't be included ]
481 must look at mkid to see how it works (for subdir usage)
482 [ right now, it doesn't. i don't see a simple fix right now ]
484 if configure.in not found, move up a directory and try again? This
485 could eliminate a common source of problems.
486 [ this is just a bad idea ]
488 * scripts are installed in $exec_prefix/bin, not $prefix/bin
490 [ the consensus on Gnits is that this isn't required.
491 doubters can work around it anyway ]
493 * make the auto-dep code crash if GNU make not in use?
494 (doesn't it already?)
496 Looked at a program called 'ezmake', which seems to do something
497 similar. The only idea there that is possibly worth stealing is using
498 globs in definitions. Also has negations. Eg in a directory with
499 files a.c, b.c and c.c, the line:
500 foo_SOURCES = *.c ~c.c
501 would be equivalent to:
502 foo_SOURCES = a.c b.c
503 Is this worth implementing?
504 [ No... it is more reliable to spell everything out. ]
506 Scan source directories and warn about missing files, eg .c/.h files
507 that aren't mentioned?
508 [ distcheck makes this less useful ]