1 Priorities for release:
2 !! documentation (eg new macros)
7 * copyrights on m4 files, aclocal output
10 * completely handle multi-":" mode for AC_CONFIG_HEADER
11 * Scan multiple input files when Makefile is generated?
12 This would provide flexibility for large projects; subsumes
13 the "Makefile.tmpl" idea
15 [ can't do this. must explain why in manual.
16 basically, solving all the problems is too hard
17 like: how to remove redundancies between generated .in files
18 instead should implement `include' directive for Makefile.am ]
20 * figure out what to pass as CONFIG_HEADER to config.status
22 * look @ jim's $(t:=foo) problem
24 * should install-foo target for a new dir (eg fooexecdir) be run via
25 install-data or install-exec? Consider examining name of variable.
27 * no mention of $(LIBTOOL) in "clean" Makefile (?)
29 * compatibility with older versions of autoconf if not --gnu
31 ================================================================
33 * only remove libtool at top level?
35 * clean up source directory by moving stuff into subdirs
37 * consider adding pkglibexecdir, maybe others?
38 requests for pkg-dirs with version included
40 Fix ansi2knr so that knr-style function decls aren't incorrectly
41 rewritten. Or, fix automake so that not all sources in a directory
42 must be ansi. Should also fix things so that a separate ._o file is
43 not needed; instead use a wrapper script.
45 Gord on pretty-printing instead of raw loops:
46 Gord> I think the best thing to do is to have make output be a valid shell
47 Gord> script, and to have it be clear exactly what is happening at a given
53 - consider adding "echo"s to installs again.
54 users can use make SHELL='sh -x' to get the full dirt
56 Avoid loops when installing; instead unroll them in automake
58 Some long-term projects:
59 * if $(FOO) is used somewhere, ensure FOO is defined, either by
60 user or by automake if possible
61 * Don't rearrange order of `include' lines relative to += assignments.
62 * Handle += assignments at all.
64 consider putting all check-* targets onto @check?
65 To support --help/--version checking?
67 take diff-n-query code from libit
70 Per> 1) Being able to build a set of non-source programs
71 Per> from source programs, without necessarily linking them together.
72 Per> I.e. one should be able to say something like:
73 Per> dummy_SOURCES=foo.c bar.c
74 Per> and automake should realize that it needs to build foo.o and bar.o.
75 Per> 2) Being intelligent about new kinds of suffixes.
77 Per> SUFFIXES = .class .java
78 Per> and a suffix rule of the form:
80 Per> then it should be able to realize it can build .class files from
81 Per> .java files, and thus be able to generate a list of
82 Per> .class files from a list of .java source files.
85 * allow a way to use "gzip --best"
86 [ Try GZIP=--best make dist ]
87 * don't assume GNU tar is "tar" (eg in distcheck)
89 * actually use acinstall program
91 !! Must fix require_file stuff. It is really gross, and I don't
92 understand it any more.
94 * error messages should print ``[info blah blah]'' command when a
95 certain part of the standards apply. saw idea in message from
96 Craig Burley. wouldn't it be really cool if compile-mode in Emacs
97 understood this convention, and you could click on such text to
98 go to the appropriate info page?
100 !! should write autoconf-style doc entries for each m4 macro
102 Jim's idea: should look for @setfilename and warn if filenames too long
104 * allow ".info" to be missing
106 * must update GNU Hello
108 ** many requests for a way to omit a file from the distribution.
109 Should be done like `!foo' or `~foo' in _SOURCES, etc.
110 Such files should be removed explicitly after the copy step!
111 Doing this requires rewriting macros before generating Makefile.in.
113 from joerg-martin schwarz:
114 -- If Makefile.am contains $(CC), $(COMPILE), $(YLWRAP), ....
115 in an explicitly written rule, you should emit the corresponding
116 Makefile variables automatically.
118 Configuring in the large:
119 * allow hierarchy of dirs to share one aclocal.m4
122 consider printing full file name of Makefile.am or configure.in when
123 giving error. This would help for very large trees with many
126 From the GNU Standards. These things could be checked, and probably
128 * Make sure that the directory into which the distribution unpacks (as
129 well as any subdirectories) are all world-writable (octal mode 777).
130 * Make sure that no file name in the distribution is more than 14
132 * Don't include any symbolic links in the distribution itself.
134 * Make sure that all the files in the distribution are world-readable.
135 ** also, check --help output and --version output. Idea from François
136 * standards no longer prohibit ANSI C. What does this imply
137 for the de-ansi-fication feature?
139 consider supporting "var+= stuff" syntax. rewrite to just var=... on
140 output. This is sometimes convenient when you want to write a
141 Makefile.am in more-or-less modular parts
143 should be able to determine what is built by looking at rules (and
144 configure.in). Then built man pages (eg) could automatically be
145 omitted from the distribution.
147 Idea from Joerg-Martin Schwarz: allow passing different -D flags to
148 different compiles. This can be done, but with the restriction that a
149 .c cannot appear in 2 different "objects" (programs/libraries)
150 compiled with different -D options (because -c and -o do not always
151 work together and parallel makes must work). This could be
152 implemented by noticing whenever a ".o" target with no rules is being
153 emitted, and adding the appropriate compilation rule as appropriate.
154 This should work with targets from Makefile.am as well as from .P
155 files, which means rewriting so that the Makefile.am contents aren't
156 copied into the output immediately. This feature is probably required
157 to fully support libtool ("grody compilation issue")
158 [ this could be probably done more directly by examining the sources
159 as we scan Makefile.am ]
161 Henrik Frystyk Nielsen says:
162 Henrik> 4) Flags like --include-deps are lost when you make changes to
163 Henrik> Makefile.am files and automake is run automatically. It would
164 Henrik> be nice to keep these flags as I now have to redo everything
166 ... what about other options here too?
168 Think about: maybe "make check" should just bomb if error occurs?
169 Then user must use "make -k check". This is probably more natural.
171 Consider: "cvs" option adds some cvs-specific rules?
173 Right now, targets generated internally (eg "install") are not
174 overridable by user code. This should probably be possible, even
175 though it isn't very important. This could be done by generating all
176 internal rules via a function call instead of just appending to
178 [ this will be harder to implement when scanning a rule like all-recursive
181 * Should be a way to have "nobuild_PROGRAMS" which aren't even built,
182 but which could be by running the magic make command.
185 * Must rewrite am_install_var. Should break into multiple functions.
186 This will allow the callers to be a little smarter.
187 * Rewrite clean targets.
188 * Must rewrite error handling code. Right now it is a real mess
189 Should fix up require_file junk at the same time
191 Things to finish libtool support:
192 * Handle grody compilation issue
193 Namely: some objects require a different compilation rule
194 * Handle install changes
195 * Handle clean changes
196 * New definition for LINK
198 djm wants ``LINKS'' variable; list of things to link together after
199 install. In BSD environment, use:
200 LINKS = from1 to1 from2 to2 ...
202 Need way to say there are no suffixes in a Makefile (Franc,ois'
203 "override" idea suffices here)
205 Check to make sure various scripts are executable (IE when looking for
208 Use recode in dist target when MAINT_CHARSET specified. Read caveats
209 in automake.in before doing this. Note the same problem used to apply
210 to the no-dependencies option; maybe it still should? Note also that
211 each Makefile.am must be rewritten at "make dist" time if
212 MAINT_CHARSET and DIST_CHARSET are not identical. NOTE: gettext must
213 arrange for all .po files not to be recoded. In the long term this
214 might be a problem (consider when some systems use Unicode but the
216 MAINT_CHARSET *must* be local to each Makefile.am, to enable
217 merged distributions.
218 DIST_CHARSET must be passed down to subdir makes during a "make dist"
220 Handle dist-zoo. Generally add more DOS support. Maybe run "doschk"
221 (why isn't this merged with "pathchk"?) when doing a dist. Do
222 whatever else François says here...
224 Add support for html via an option. Use texi2html. Use
225 "html_TEXINFOS", and htmldir = .../html. Include html files in
226 distribution. Also allow "html_DATA", for raw .html files.
227 [ when will texinfo directly support html? ]
229 uninstall and pkg-dirs should rm -rf the dir.
231 a potential bug: configure puts "blah.o" into LIBOBJS, thus implying
232 these files can't be de-ansified. Not a problem?
233 [ fix by using ansi2knr wrapper program ]
235 In general most .am files should be merged into automake. For
236 instance all the "clean" targets could be merged by keeping lists of
237 things to be removed. This would be a lot nicer looking. Note that
238 the install targets probably should not be merged; it is sometimes
239 useful to only install a small part.
242 * Order rules sensibly
243 * Ensure every line has a purpose. Omit unused stuff
244 * Eliminate extraneous rules when possible (eg 'install-am' stuff)
245 * Make sure vertical spacing is correct
246 Omit program transform vars from header if no program installed. This
247 is currently pretty hard to do. (But with beautification code it
248 would probably be easy)
251 * It would be nice to automatically support using bison's better features
252 to rename the output files. This requires autoconf support
253 * Consider supporting syntax from autoconf "derived:source", eg:
255 for yacc and lex source
256 * allow generating c++ source from lex? Eg using flex?
258 Multi-language support:
259 * should have mapping of file extensions to languages
260 * should automatically handle the linking issue (special-case C++)
261 * must get compile rules for various languages; FORTRAN probably
262 most important unimplemented language
263 This should be integrated in some way with Per's idea.
264 Eg .f.o rules should be recognized & auto-handled in _SOURCES
265 That way any random language can be treated with C/C++ on a first-class
268 It might be cool to generate .texi dependencies by grepping for
269 @include. (If done, it should be done the same way C dependencies are
272 It would be good to check some parts of GNU standards. Already check
273 for install-sh and mkinstalldirs. What else is required to be in
274 package by GNU standards or by automake?
275 Some things for --strictness=gnits:
276 * "cd $(foo); something" is an error in a rule. Should be:
277 "cd $(foo) && something"
278 * Look for 'ln -s' and warn about using $(LN) and AC_PROG_LN_S
279 * Look for $(LN) and require AC_PROG_LN_S
281 Auto-distribute "ChangeLog.[0-9]+"? "ChangeLog.[a-z]+"?
283 Internationalize. [ gettext doesn't have the necessary machinery yet ]
284 am_error should use printf-style arguments (for eventual gettext scheme)
286 François says the ordering of files in a distribution should be as follows:
290 I agree, but I don't see how to implement this yet.
291 It might be easier if "derived files" is limited to those that
292 Automake itself knows about, eg output of yacc.
294 Check all source files to make sure that FSF address is up-to-date.
295 --gnits or --gnu only.
297 Merge each -vars.am file with corresponding ".am" file. Can do this
298 because of changes to &file_contents.
300 Should libexec programs have the name transform done on them?
302 Order the output rules sensibly, so FOO_SOURCES and FOO_OBJECTS are
303 together and rules are in the usual order.
305 Make the output minimal: only output definitions for variables that
308 Look at dist's jmake for ideas. dist is the name of the distribution
309 including Metaconfig. Perl uses it.
311 Should handle directory hierarchies deeper than 2. Right now there is
312 some support for this. Here are some of the issues:
313 * Should handle AC_CONFIG_SUBDIRS, ie must handle configure.in in subdirs
314 * can do this by looking at subdirs, seeing configure.in
315 and auto-running Automake there
317 . Consider supporting guile-style PLUGIN directories automatically?
320 David> To avoid comments like the one about subdirs getting buried in
321 David> the middle of a Makefile.in, how about pushing comments that
322 David> start with ### to the top of the Makefile.in (in order)? Sort
323 David> of like how Autoconf uses diversions to force initialization
324 David> code to the top of configure.
327 Karl> 2) Your Makefile variable names are generally uppercase, but GNU
328 Karl> generally uses lowercase. Not that it matters :-).
330 ================================================================
334 probably should put each group of m4 files into a subdir owned by the
335 containing application.
337 ================================================================
341 rationale for avoiding
342 make CFLAGS="$CFLAGS" ...
345 a package that installs its own aclocal macros
351 write example of using automake with dejagnu
352 follow calc example in dejagnu docs
354 document which variables are actually scanned and which are not.
356 Document customary ordering of Makefile.am. From François.
358 Should include extended version of diagram from Autoconf (suggested by
361 Make a definition of the term "source"
363 document how to use Automake with CVS. Idea from Mark Galassi. Also
364 include Greg Woods' more sophisticated "cvs-dist" target.
366 document rebuilding configure. CONFIGURE_DEPENDENCIES
368 _DEPENDENCIES -vs- _LIBADD/_LDADD; the tradeoffs
370 -- must document all variables that are supposed
371 to be public knowledge
373 automake must be run in each directory with a configure.in
374 This is insufficiently clear
376 must document the targets required for integration with
377 non-automake-using subdirs
379 use of (eg) EXTRA_PROGRAMS is not very clear right now
381 document EXTRA_foo_SOURCES
383 document why EXTRA_* vars must be statically knowable
385 document the "make SHELL='/bin/sh -x'" trick for debugging
387 section on relationship to GNU make
389 LDFLAGS; maybe why it is different from LIBADD/LDADD
391 document that dependencies:
393 * are system-dependent
395 ================================================================
397 Things to do for autoconf:
399 * patch autoreconf to run automake and aclocal. I've done this but it is
400 not really available. It can't be made available until automake
401 is officially released
403 ================================================================
407 * Should support standalone library along with subdir library in same
408 Makefile.am. Maybe: turn off "standalone" mode if library's Makefile.am
409 is not only one specd? [ add an option for this ]
411 ================================================================
415 Would it be useful to integrate in some way with the Debian package
416 building utility? Must check. maybe it would be possible to deal
417 with all the different package utilities somehow. Lately I've been
418 hearing good things about the RedHat packaging utilities. Why are
419 there so many of these? Are they fun to write or something?
420 The RedHat package utility is called RPM; see
421 ftp://ftp.redhat.com/pub/code/rpm
422 It actually has problems, like no configure script and no documentation.
424 For Cygnus it would probably be good to be able to handle the native
425 package utility on each platform. There are probably 3 or 4 of these
426 (sysv, solaris?, aix?)
428 tcl/unix/Makefile.in has some code to generate a Solaris package.
430 Automake probably can't do all of this on its own. A new tool might
433 ================================================================
435 A tool to guess what the local Makefile.am should look like:
436 (see Gord's Maint program!)
438 * Probably integrate with autoscan
439 * Use various simple rules to determine what to do:
440 * get name of top directory, sans version info
441 * search for .c files with 'main' in them
442 * if in main.c, use directory name for program
443 * if in more than one, generate multiple programs
444 * if not found, generate a library named after directory
445 * order subdir searches correctly: lib first, src last
446 * assume 'testsuite' dir means we are using dejagnu
447 * maybe be smart about reading existing Makefile.am, so tool
448 can be run for incremental changes? You could imagine:
451 autoproject --incremental
453 ================================================================
455 Stuff NOT to do, and why:
457 consider auto-including any file that matches "*.in".
458 [ no: po/Makefile.in shouldn't be included ]
460 must look at mkid to see how it works (for subdir usage)
461 [ right now, it doesn't. i don't see a simple fix right now ]
463 if configure.in not found, move up a directory and try again? This
464 could eliminate a common source of problems.
465 [ this is just a bad idea ]
467 * scripts are installed in $exec_prefix/bin, not $prefix/bin
469 [ the consensus on Gnits is that this isn't required.
470 doubters can work around it anyway ]
472 * make the auto-dep code crash if GNU make not in use?
473 (doesn't it already?)
475 Looked at a program called 'ezmake', which seems to do something
476 similar. The only idea there that is possibly worth stealing is using
477 globs in definitions. Also has negations. Eg in a directory with
478 files a.c, b.c and c.c, the line:
479 foo_SOURCES = *.c ~c.c
480 would be equivalent to:
481 foo_SOURCES = a.c b.c
482 Is this worth implementing?
483 [ No... it is more reliable to spell everything out. ]
485 Scan source directories and warn about missing files, eg .c/.h files
486 that aren't mentioned?
487 [ distcheck makes this less useful ]