1 * update text in missing
4 ** make sure every variable that is used is also defined
6 * make sure `missing' defines are generated
7 * if no AM_INIT_AUTOMAKE, then don't handle `missing' stuff.
10 * allow 'cygnus' in AUTOMAKE_OPTIONS
14 * NORMAL_INSTALL / NORMAL_UNINSTALL -vs- recursive rules
15 [ requires changes to the standard ]
17 * if foo.y is a source, foo.h isn't auto-distributed?
19 * dependency tracking doesn't work well when a file is removed
20 the new code to track header dependencies exacerbates this
22 it would probably be better to use "gcc -MD" and move the .d
23 file into the .deps directory. That is, create the dependencies
24 as a side effect of compilation
25 This still won't solve the file-deletion problem
27 * copyrights on m4 files, aclocal output
29 * is there a way to add a directory and then have "make" do all the
32 * put standards.texi into distribution
35 * should not put texiname_TEXINFOS into distribution
36 should rename this macro anyway, to foo_texi_DEPENDENCIES
38 * *all* installed scripts should support --version, --help
40 * have aclocal diagnose unrecognized AM_ macros
42 For now I guess I'll just have automake give an error if it encounters
43 non-C source in a libtool library specification.
45 * must split $obj into two parts: one for libtool and one for
46 deansification. Otherwise .S files will be deansified!
48 * if program has the same name as a target, do something sensible:
49 - if the target is internal, rename it
50 - if the target is mandated (eg, "info"), tell the user
51 consider auto-modifying the program name to work around this
53 * should separate actual options from strictness levels
54 strictness should only cover requirements
55 You should be able to pick and choose options
57 should clean up texinfos.am; one rule is repeated 3 times, but
60 should always use perl -w
62 rewrite in guile (RMS request)
63 at the same time, consider adding a GUI
64 could use the same parsing code for the GUI and the standalone version
65 that means figuring out a better representation of internal state
67 having just one Makefile for a project would give a big speed increase
68 for a project with many directories, eg glibc. ideally (?) you'd
69 still be able to have a Makefile.am in each directory somehow; this
70 might make editing conceptually easier.
74 * put parser.h into distribution if "yacc -d" is used
76 * only remove libtool at top level?
78 * clean up source directory by moving stuff into subdirs
80 * consider adding pkglibexecdir, maybe others?
81 requests for pkg-dirs with version included
86 Avoid loops when installing; instead unroll them in automake
89 * completely handle multi-":" mode for AC_CONFIG_HEADER
90 * Scan multiple input files when Makefile is generated?
91 This would provide flexibility for large projects; subsumes
92 the "Makefile.tmpl" idea
94 [ can't do this. must explain why in manual.
95 basically, solving all the problems is too hard
96 like: how to remove redundancies between generated .in files
97 instead should implement `include' directive for Makefile.am ]
98 * for multi-":" mode and AC_OUTPUT, it might be good to pick the
99 first input file that has a corresponding .am file.
101 Some long-term projects:
102 * if $(FOO) is used somewhere, ensure FOO is defined, either by
103 user or by automake if possible
104 * Don't rearrange order of `include' lines relative to += assignments.
105 * Handle += assignments at all.
106 * Handle `include' lines by scanning other files, and adding
107 to Makefile.in dependency
109 consider putting all check-* targets onto @check?
110 To support --help/--version checking?
112 take diff-n-query code from libit
115 Per> 1) Being able to build a set of non-source programs
116 Per> from source programs, without necessarily linking them together.
117 Per> I.e. one should be able to say something like:
118 Per> dummy_SOURCES=foo.c bar.c
119 Per> and automake should realize that it needs to build foo.o and bar.o.
120 Per> 2) Being intelligent about new kinds of suffixes.
122 Per> SUFFIXES = .class .java
123 Per> and a suffix rule of the form:
125 Per> then it should be able to realize it can build .class files from
126 Per> .java files, and thus be able to generate a list of
127 Per> .class files from a list of .java source files.
129 !! Must fix require_file stuff. It is really gross, and I don't
130 understand it any more.
132 * error messages should print ``[info blah blah]'' command when a
133 certain part of the standards apply. saw idea in message from
134 Craig Burley. wouldn't it be really cool if compile-mode in Emacs
135 understood this convention, and you could click on such text to
136 go to the appropriate info page?
138 Jim's idea: should look for @setfilename and warn if filenames too long
141 ** many requests for a way to omit a file from the distribution.
142 Should be done like `!foo' or `~foo' in _SOURCES, etc.
143 Such files should be removed explicitly after the copy step!
144 Doing this requires rewriting macros before generating Makefile.in.
146 from joerg-martin schwarz:
147 -- If Makefile.am contains $(CC), $(COMPILE), $(YLWRAP), ....
148 in an explicitly written rule, you should emit the corresponding
149 Makefile variables automatically.
151 Configuring in the large:
152 * allow hierarchy of dirs to share one aclocal.m4
155 consider printing full file name of Makefile.am or configure.in when
156 giving error. This would help for very large trees with many
159 From the GNU Standards. These things could be checked, and probably
161 * Make sure that the directory into which the distribution unpacks (as
162 well as any subdirectories) are all world-writable (octal mode 777).
163 * Make sure that no file name in the distribution is more than 14
165 * Don't include any symbolic links in the distribution itself.
167 * Make sure that all the files in the distribution are world-readable.
168 ** also, check --help output and --version output. Idea from François
169 * standards no longer prohibit ANSI C. What does this imply
170 for the de-ansi-fication feature?
172 consider supporting "var+= stuff" syntax. rewrite to just var=... on
173 output. This is sometimes convenient when you want to write a
174 Makefile.am in more-or-less modular parts
176 should be able to determine what is built by looking at rules (and
177 configure.in). Then built man pages (eg) could automatically be
178 omitted from the distribution.
180 Idea from Joerg-Martin Schwarz: allow passing different -D flags to
181 different compiles. This can be done, but with the restriction that a
182 .c cannot appear in 2 different "objects" (programs/libraries)
183 compiled with different -D options (because -c and -o do not always
184 work together and parallel makes must work). This could be
185 implemented by noticing whenever a ".o" target with no rules is being
186 emitted, and adding the appropriate compilation rule as appropriate.
187 This should work with targets from Makefile.am as well as from .P
188 files, which means rewriting so that the Makefile.am contents aren't
189 copied into the output immediately.
190 [ this could be probably done more directly by examining the sources
191 as we scan Makefile.am ]
193 Henrik Frystyk Nielsen says:
194 Henrik> 4) Flags like --include-deps are lost when you make changes to
195 Henrik> Makefile.am files and automake is run automatically. It would
196 Henrik> be nice to keep these flags as I now have to redo everything
198 ... what about other options here too?
200 Think about: maybe "make check" should just bomb if error occurs?
201 Then user must use "make -k check". This is probably more natural.
203 Consider: "cvs" option adds some cvs-specific rules?
205 Right now, targets generated internally (eg "install") are not
206 overridable by user code. This should probably be possible, even
207 though it isn't very important. This could be done by generating all
208 internal rules via a function call instead of just appending to
210 [ this will be harder to implement when scanning a rule like all-recursive
213 * Should be a way to have "nobuild_PROGRAMS" which aren't even built,
214 but which could be by running the magic make command.
217 * Must rewrite am_install_var. Should break into multiple functions.
218 This will allow the callers to be a little smarter.
219 * Rewrite clean targets.
220 * Must rewrite error handling code. Right now it is a real mess
221 Should fix up require_file junk at the same time
223 djm wants ``LINKS'' variable; list of things to link together after
224 install. In BSD environment, use:
225 LINKS = from1 to1 from2 to2 ...
227 Need way to say there are no suffixes in a Makefile (Franc,ois'
228 "override" idea suffices here)
230 Check to make sure various scripts are executable (IE when looking for
233 Use recode in dist target when MAINT_CHARSET specified. Read caveats
234 in automake.in before doing this. Note the same problem used to apply
235 to the no-dependencies option; maybe it still should? Note also that
236 each Makefile.am must be rewritten at "make dist" time if
237 MAINT_CHARSET and DIST_CHARSET are not identical. NOTE: gettext must
238 arrange for all .po files not to be recoded. In the long term this
239 might be a problem (consider when some systems use Unicode but the
241 MAINT_CHARSET *must* be local to each Makefile.am, to enable
242 merged distributions.
243 DIST_CHARSET must be passed down to subdir makes during a "make dist"
245 Handle dist-zoo. Generally add more DOS support. Maybe run "doschk"
246 (why isn't this merged with "pathchk"?) when doing a dist. Do
247 whatever else François says here...
249 Add support for html via an option. Use texi2html. Use
250 "html_TEXINFOS", and htmldir = .../html. Include html files in
251 distribution. Also allow "html_DATA", for raw .html files.
252 [ when will texinfo directly support html? ]
254 uninstall and pkg-dirs should rm -rf the dir.
256 a potential bug: configure puts "blah.o" into LIBOBJS, thus implying
257 these files can't be de-ansified. Not a problem?
258 [ fix by using ansi2knr wrapper program ]
260 In general most .am files should be merged into automake. For
261 instance all the "clean" targets could be merged by keeping lists of
262 things to be removed. This would be a lot nicer looking. Note that
263 the install targets probably should not be merged; it is sometimes
264 useful to only install a small part.
267 * Order rules sensibly
268 * Ensure every line has a purpose. Omit unused stuff
269 * Eliminate extraneous rules when possible (eg 'install-am' stuff)
270 * Make sure vertical spacing is correct
271 Omit program transform vars from header if no program installed. This
272 is currently pretty hard to do. (But with beautification code it
273 would probably be easy)
276 * It would be nice to automatically support using bison's better features
277 to rename the output files. This requires autoconf support
278 * Consider supporting syntax from autoconf "derived:source", eg:
280 for yacc and lex source
281 * what if you use flex and the option to avoid -lfl?
284 Multi-language support:
285 * should have mapping of file extensions to languages
286 * should automatically handle the linking issue (special-case C++)
287 * must get compile rules for various languages; FORTRAN probably
288 most important unimplemented language
289 This should be integrated in some way with Per's idea.
290 Eg .f.o rules should be recognized & auto-handled in _SOURCES
291 That way any random language can be treated with C/C++ on a first-class
294 It might be cool to generate .texi dependencies by grepping for
295 @include. (If done, it should be done the same way C dependencies are
298 It would be good to check some parts of GNU standards. Already check
299 for install-sh and mkinstalldirs. What else is required to be in
300 package by GNU standards or by automake?
301 Some things for --strictness=gnits:
302 * "cd $(foo); something" is an error in a rule. Should be:
303 "cd $(foo) && something"
304 * Look for 'ln -s' and warn about using $(LN) and AC_PROG_LN_S
305 * Look for $(LN) and require AC_PROG_LN_S
307 Auto-distribute "ChangeLog.[0-9]+"? "ChangeLog.[a-z]+"?
309 Internationalize. [ gettext doesn't have the necessary machinery yet ]
310 am_error should use printf-style arguments (for eventual gettext scheme)
312 François says the ordering of files in a distribution should be as follows:
316 I agree, but I don't see how to implement this yet.
317 It might be easier if "derived files" is limited to those that
318 Automake itself knows about, eg output of yacc.
320 Check all source files to make sure that FSF address is up-to-date.
321 --gnits or --gnu only.
323 Merge each -vars.am file with corresponding ".am" file. Can do this
324 because of changes to &file_contents.
326 Should libexec programs have the name transform done on them?
328 Order the output rules sensibly, so FOO_SOURCES and FOO_OBJECTS are
329 together and rules are in the usual order.
331 Make the output minimal: only output definitions for variables that
335 David> To avoid comments like the one about subdirs getting buried in
336 David> the middle of a Makefile.in, how about pushing comments that
337 David> start with ### to the top of the Makefile.in (in order)? Sort
338 David> of like how Autoconf uses diversions to force initialization
339 David> code to the top of configure.
342 Karl> 2) Your Makefile variable names are generally uppercase, but GNU
343 Karl> generally uses lowercase. Not that it matters :-).
345 ================================================================
349 probably should put each group of m4 files into a subdir owned by the
350 containing application.
352 ================================================================
358 how to use the generated makefiles
361 - NORMAL_INSTALL junk
363 what goes in AC_CONFIG_AUX_DIR
365 multi-":" mode in AC_OUTPUT -- automake only looks at the first file
366 also a note on how a .am file is found in this case
368 rationale for avoiding
369 make CFLAGS="$CFLAGS" ...
372 a package that installs its own aclocal macros
374 write example of using automake with dejagnu
375 follow calc example in dejagnu docs
377 document which variables are actually scanned and which are not.
379 Document customary ordering of Makefile.am. From François.
381 Should include extended version of diagram from Autoconf (suggested by
384 Make a definition of the term "source"
386 document how to use Automake with CVS. Idea from Mark Galassi. Also
387 include Greg Woods' more sophisticated "cvs-dist" target.
389 document rebuilding configure. CONFIGURE_DEPENDENCIES
391 -- must document all variables that are supposed
392 to be public knowledge
394 automake must be run in each directory with a configure.in
395 This is insufficiently clear
397 must document the targets required for integration with
398 non-automake-using subdirs
400 document the "make SHELL='/bin/sh -x'" trick for debugging
402 section on relationship to GNU make
406 ================================================================
408 Things to do for autoconf:
410 * patch autoreconf to run automake and aclocal. I've done this but it is
411 not really available. It can't be made available until automake
412 is officially released
414 ================================================================
418 * Should support standalone library along with subdir library in same
419 Makefile.am. Maybe: turn off "standalone" mode if library's Makefile.am
420 is not only one specd? [ add an option for this ]
422 ================================================================
426 Would it be useful to integrate in some way with the Debian package
427 building utility? Must check. maybe it would be possible to deal
428 with all the different package utilities somehow. Lately I've been
429 hearing good things about the RedHat packaging utilities. Why are
430 there so many of these? Are they fun to write or something?
431 The RedHat package utility is called RPM; see
432 ftp://ftp.redhat.com/pub/code/rpm
433 It actually has problems, like no configure script and no documentation.
435 For Cygnus it would probably be good to be able to handle the native
436 package utility on each platform. There are probably 3 or 4 of these
437 (sysv, solaris?, aix?)
439 tcl/unix/Makefile.in has some code to generate a Solaris package.
441 Automake probably can't do all of this on its own. A new tool might
444 I have some notes from a Debian developer on how the integration
447 ================================================================
449 A tool to guess what the local Makefile.am should look like:
450 (see Gord's Maint program!)
452 * Probably integrate with autoscan
453 * Use various simple rules to determine what to do:
454 * get name of top directory, sans version info
455 * search for .c files with 'main' in them
456 * if in main.c, use directory name for program
457 * if in more than one, generate multiple programs
458 * if not found, generate a library named after directory
459 * order subdir searches correctly: lib first, src last
460 * assume 'testsuite' dir means we are using dejagnu
461 * maybe be smart about reading existing Makefile.am, so tool
462 can be run for incremental changes? You could imagine:
465 autoproject --incremental
467 ================================================================
469 Stuff NOT to do, and why:
471 consider auto-including any file that matches "*.in".
472 [ no: po/Makefile.in shouldn't be included ]
474 must look at mkid to see how it works (for subdir usage)
475 [ right now, it doesn't. i don't see a simple fix right now ]
477 if configure.in not found, move up a directory and try again? This
478 could eliminate a common source of problems.
479 [ this is just a bad idea ]
481 * scripts are installed in $exec_prefix/bin, not $prefix/bin
483 [ the consensus on Gnits is that this isn't required.
484 doubters can work around it anyway ]
486 * make the auto-dep code crash if GNU make not in use?
487 (doesn't it already?)
489 Looked at a program called 'ezmake', which seems to do something
490 similar. The only idea there that is possibly worth stealing is using
491 globs in definitions. Also has negations. Eg in a directory with
492 files a.c, b.c and c.c, the line:
493 foo_SOURCES = *.c ~c.c
494 would be equivalent to:
495 foo_SOURCES = a.c b.c
496 Is this worth implementing?
497 [ No... it is more reliable to spell everything out. ]
499 Scan source directories and warn about missing files, eg .c/.h files
500 that aren't mentioned?
501 [ distcheck makes this less useful ]