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
14 Figuring out when to update the sub-parts is hard, maybe
15 * figure out what to pass as CONFIG_HEADER to config.status
17 * look @ jim's $(t:=foo) problem
19 * should install-foo target for a new dir (eg fooexecdir) be run via
20 install-data or install-exec? Consider examining name of variable.
22 ================================================================
24 * only remove libtool at top level?
26 * clean up source directory by moving stuff into subdirs
28 * consider adding pkglibexecdir, maybe others?
29 requests for pkg-dirs with version included
31 Fix ansi2knr so that knr-style function decls aren't incorrectly
32 rewritten. Or, fix automake so that not all sources in a directory
33 must be ansi. Should also fix things so that a separate ._o file is
34 not needed; instead use a wrapper script.
36 Gord on pretty-printing instead of raw loops:
37 Gord> I think the best thing to do is to have make output be a valid shell
38 Gord> script, and to have it be clear exactly what is happening at a given
44 - consider adding "echo"s to installs again.
45 users can use make SHELL='sh -x' to get the full dirt
47 Avoid loops when installing; instead unroll them in automake
49 Some long-term projects:
50 * if $(FOO) is used somewhere, ensure FOO is defined, either by
51 user or by automake if possible
52 * Don't rearrange order of `include' lines relative to += assignments.
53 * Handle += assignments at all.
55 consider putting all check-* targets onto @check?
56 To support --help/--version checking?
58 take diff-n-query code from libit
61 Per> 1) Being able to build a set of non-source programs
62 Per> from source programs, without necessarily linking them together.
63 Per> I.e. one should be able to say something like:
64 Per> dummy_SOURCES=foo.c bar.c
65 Per> and automake should realize that it needs to build foo.o and bar.o.
66 Per> 2) Being intelligent about new kinds of suffixes.
68 Per> SUFFIXES = .class .java
69 Per> and a suffix rule of the form:
71 Per> then it should be able to realize it can build .class files from
72 Per> .java files, and thus be able to generate a list of
73 Per> .class files from a list of .java source files.
76 * allow a way to use "gzip --best"
77 [ Try GZIP=--best make dist ]
78 * don't assume GNU tar is "tar" (eg in distcheck)
80 * actually use acinstall program
82 !! Must fix require_file stuff. It is really gross, and I don't
83 understand it any more.
85 * error messages should print ``[info blah blah]'' command when a
86 certain part of the standards apply. saw idea in message from
87 Craig Burley. wouldn't it be really cool if compile-mode in Emacs
88 understood this convention, and you could click on such text to
89 go to the appropriate info page?
91 !! should write autoconf-style doc entries for each m4 macro
93 Jim's idea: should look for @setfilename and warn if filenames too long
95 * allow ".info" to be missing
97 * must update GNU Hello
99 ** many requests for a way to omit a file from the distribution.
100 Should be done like `!foo' or `~foo' in _SOURCES, etc.
101 Such files should be removed explicitly after the copy step!
102 Doing this requires rewriting macros before generating Makefile.in.
104 from joerg-martin schwarz:
105 -- If Makefile.am contains $(CC), $(COMPILE), $(YLWRAP), ....
106 in an explicitly written rule, you should emit the corresponding
107 Makefile variables automatically.
109 Configuring in the large:
110 * allow hierarchy of dirs to share one aclocal.m4
113 consider printing full file name of Makefile.am or configure.in when
114 giving error. This would help for very large trees with many
117 From the GNU Standards. These things could be checked, and probably
119 * Make sure that the directory into which the distribution unpacks (as
120 well as any subdirectories) are all world-writable (octal mode 777).
121 * Make sure that no file name in the distribution is more than 14
123 * Don't include any symbolic links in the distribution itself.
125 * Make sure that all the files in the distribution are world-readable.
126 ** also, check --help output and --version output. Idea from François
127 * standards no longer prohibit ANSI C. What does this imply
128 for the de-ansi-fication feature?
130 consider supporting "var+= stuff" syntax. rewrite to just var=... on
131 output. This is sometimes convenient when you want to write a
132 Makefile.am in more-or-less modular parts
134 should be able to determine what is built by looking at rules (and
135 configure.in). Then built man pages (eg) could automatically be
136 omitted from the distribution.
138 Idea from Joerg-Martin Schwarz: allow passing different -D flags to
139 different compiles. This can be done, but with the restriction that a
140 .c cannot appear in 2 different "objects" (programs/libraries)
141 compiled with different -D options (because -c and -o do not always
142 work together and parallel makes must work). This could be
143 implemented by noticing whenever a ".o" target with no rules is being
144 emitted, and adding the appropriate compilation rule as appropriate.
145 This should work with targets from Makefile.am as well as from .P
146 files, which means rewriting so that the Makefile.am contents aren't
147 copied into the output immediately. This feature is probably required
148 to fully support libtool ("grody compilation issue")
149 [ this could be probably done more directly by examining the sources
150 as we scan Makefile.am ]
152 Henrik Frystyk Nielsen says:
153 Henrik> 4) Flags like --include-deps are lost when you make changes to
154 Henrik> Makefile.am files and automake is run automatically. It would
155 Henrik> be nice to keep these flags as I now have to redo everything
157 ... what about other options here too?
159 Think about: maybe "make check" should just bomb if error occurs?
160 Then user must use "make -k check". This is probably more natural.
162 Consider: "cvs" option adds some cvs-specific rules?
164 Right now, targets generated internally (eg "install") are not
165 overridable by user code. This should probably be possible, even
166 though it isn't very important. This could be done by generating all
167 internal rules via a function call instead of just appending to
169 [ this will be harder to implement when scanning a rule like all-recursive
172 * Should be a way to have "nobuild_PROGRAMS" which aren't even built,
173 but which could be by running the magic make command.
176 * Must rewrite am_install_var. Should break into multiple functions.
177 This will allow the callers to be a little smarter.
178 * Rewrite clean targets.
179 * Must rewrite error handling code. Right now it is a real mess
180 Should fix up require_file junk at the same time
182 Things to finish libtool support:
183 * Handle grody compilation issue
184 Namely: some objects require a different compilation rule
185 * Handle install changes
186 * Handle clean changes
187 * New definition for LINK
189 djm wants ``LINKS'' variable; list of things to link together after
190 install. In BSD environment, use:
191 LINKS = from1 to1 from2 to2 ...
193 Need way to say there are no suffixes in a Makefile (Franc,ois'
194 "override" idea suffices here)
196 Check to make sure various scripts are executable (IE when looking for
199 Use recode in dist target when MAINT_CHARSET specified. Read caveats
200 in automake.in before doing this. Note the same problem used to apply
201 to the no-dependencies option; maybe it still should? Note also that
202 each Makefile.am must be rewritten at "make dist" time if
203 MAINT_CHARSET and DIST_CHARSET are not identical. NOTE: gettext must
204 arrange for all .po files not to be recoded. In the long term this
205 might be a problem (consider when some systems use Unicode but the
207 MAINT_CHARSET *must* be local to each Makefile.am, to enable
208 merged distributions.
209 DIST_CHARSET must be passed down to subdir makes during a "make dist"
211 Handle dist-zoo. Generally add more DOS support. Maybe run "doschk"
212 (why isn't this merged with "pathchk"?) when doing a dist. Do
213 whatever else François says here...
215 Add support for html via an option. Use texi2html. Use
216 "html_TEXINFOS", and htmldir = .../html. Include html files in
217 distribution. Also allow "html_DATA", for raw .html files.
218 [ when will texinfo directly support html? ]
220 uninstall and pkg-dirs should rm -rf the dir.
222 a potential bug: configure puts "blah.o" into LIBOBJS, thus implying
223 these files can't be de-ansified. Not a problem?
224 [ fix by using ansi2knr wrapper program ]
226 In general most .am files should be merged into automake. For
227 instance all the "clean" targets could be merged by keeping lists of
228 things to be removed. This would be a lot nicer looking. Note that
229 the install targets probably should not be merged; it is sometimes
230 useful to only install a small part.
233 * Order rules sensibly
234 * Ensure every line has a purpose. Omit unused stuff
235 * Eliminate extraneous rules when possible (eg 'install-am' stuff)
236 * Make sure vertical spacing is correct
237 Omit program transform vars from header if no program installed. This
238 is currently pretty hard to do. (But with beautification code it
239 would probably be easy)
242 * It would be nice to automatically support using bison's better features
243 to rename the output files. This requires autoconf support
244 * Consider supporting syntax from autoconf "derived:source", eg:
246 for yacc and lex source
247 * allow generating c++ source from lex? Eg using flex?
249 Multi-language support:
250 * should have mapping of file extensions to languages
251 * should automatically handle the linking issue (special-case C++)
252 * must get compile rules for various languages; FORTRAN probably
253 most important unimplemented language
254 This should be integrated in some way with Per's idea.
255 Eg .f.o rules should be recognized & auto-handled in _SOURCES
256 That way any random language can be treated with C/C++ on a first-class
259 It might be cool to generate .texi dependencies by grepping for
260 @include. (If done, it should be done the same way C dependencies are
263 It would be good to check some parts of GNU standards. Already check
264 for install-sh and mkinstalldirs. What else is required to be in
265 package by GNU standards or by automake?
266 Some things for --strictness=gnits:
267 * "cd $(foo); something" is an error in a rule. Should be:
268 "cd $(foo) && something"
269 * Look for 'ln -s' and warn about using $(LN) and AC_PROG_LN_S
270 * Look for $(LN) and require AC_PROG_LN_S
272 Auto-distribute "ChangeLog.[0-9]+"? "ChangeLog.[a-z]+"?
274 Internationalize. [ gettext doesn't have the necessary machinery yet ]
275 am_error should use printf-style arguments (for eventual gettext scheme)
277 François says the ordering of files in a distribution should be as follows:
281 I agree, but I don't see how to implement this yet.
282 It might be easier if "derived files" is limited to those that
283 Automake itself knows about, eg output of yacc.
285 Check all source files to make sure that FSF address is up-to-date.
286 --gnits or --gnu only.
288 Merge each -vars.am file with corresponding ".am" file. Can do this
289 because of changes to &file_contents.
291 Should libexec programs have the name transform done on them?
293 Order the output rules sensibly, so FOO_SOURCES and FOO_OBJECTS are
294 together and rules are in the usual order.
296 Make the output minimal: only output definitions for variables that
299 Look at dist's jmake for ideas. dist is the name of the distribution
300 including Metaconfig. Perl uses it.
302 Should handle directory hierarchies deeper than 2. Right now there is
303 some support for this. Here are some of the issues:
304 * Should handle AC_CONFIG_SUBDIRS, ie must handle configure.in in subdirs
305 * can do this by looking at subdirs, seeing configure.in
306 and auto-running Automake there
308 . Consider supporting guile-style PLUGIN directories automatically?
311 David> To avoid comments like the one about subdirs getting buried in
312 David> the middle of a Makefile.in, how about pushing comments that
313 David> start with ### to the top of the Makefile.in (in order)? Sort
314 David> of like how Autoconf uses diversions to force initialization
315 David> code to the top of configure.
318 Karl> 2) Your Makefile variable names are generally uppercase, but GNU
319 Karl> generally uses lowercase. Not that it matters :-).
321 ================================================================
325 probably should put each group of m4 files into a subdir owned by the
326 containing application.
328 ================================================================
332 rationale for avoiding
333 make CFLAGS="$CFLAGS" ...
336 a package that installs its own aclocal macros
342 write example of using automake with dejagnu
343 follow calc example in dejagnu docs
345 document which variables are actually scanned and which are not.
347 Document customary ordering of Makefile.am. From François.
349 Should include extended version of diagram from Autoconf (suggested by
352 Make a definition of the term "source"
354 document how to use Automake with CVS. Idea from Mark Galassi. Also
355 include Greg Woods' more sophisticated "cvs-dist" target.
357 document rebuilding configure. CONFIGURE_DEPENDENCIES
359 _DEPENDENCIES -vs- _LIBADD/_LDADD; the tradeoffs
361 -- must document all variables that are supposed
362 to be public knowledge
364 automake must be run in each directory with a configure.in
365 This is insufficiently clear
367 must document the targets required for integration with
368 non-automake-using subdirs
370 use of (eg) EXTRA_PROGRAMS is not very clear right now
372 document EXTRA_foo_SOURCES
374 document why EXTRA_* vars must be statically knowable
376 document the "make SHELL='/bin/sh -x'" trick for debugging
378 section on relationship to GNU make
380 LDFLAGS; maybe why it is different from LIBADD/LDADD
382 document that dependencies:
384 * are system-dependent
386 ================================================================
388 Things to do for autoconf:
390 * patch autoreconf to run automake and aclocal. I've done this but it is
391 not really available. It can't be made available until automake
392 is officially released
394 ================================================================
398 * Should support standalone library along with subdir library in same
399 Makefile.am. Maybe: turn off "standalone" mode if library's Makefile.am
400 is not only one specd? [ add an option for this ]
402 ================================================================
406 Would it be useful to integrate in some way with the Debian package
407 building utility? Must check. maybe it would be possible to deal
408 with all the different package utilities somehow. Lately I've been
409 hearing good things about the RedHat packaging utilities. Why are
410 there so many of these? Are they fun to write or something?
411 The RedHat package utility is called RPM; see
412 ftp://ftp.redhat.com/pub/code/rpm
413 It actually has problems, like no configure script and no documentation.
415 For Cygnus it would probably be good to be able to handle the native
416 package utility on each platform. There are probably 3 or 4 of these
417 (sysv, solaris?, aix?)
419 tcl/unix/Makefile.in has some code to generate a Solaris package.
421 Automake probably can't do all of this on its own. A new tool might
424 ================================================================
426 A tool to guess what the local Makefile.am should look like:
427 (see Gord's Maint program!)
429 * Probably integrate with autoscan
430 * Use various simple rules to determine what to do:
431 * get name of top directory, sans version info
432 * search for .c files with 'main' in them
433 * if in main.c, use directory name for program
434 * if in more than one, generate multiple programs
435 * if not found, generate a library named after directory
436 * order subdir searches correctly: lib first, src last
437 * assume 'testsuite' dir means we are using dejagnu
438 * maybe be smart about reading existing Makefile.am, so tool
439 can be run for incremental changes? You could imagine:
442 autoproject --incremental
444 ================================================================
446 Stuff NOT to do, and why:
448 consider auto-including any file that matches "*.in".
449 [ no: po/Makefile.in shouldn't be included ]
451 must look at mkid to see how it works (for subdir usage)
452 [ right now, it doesn't. i don't see a simple fix right now ]
454 if configure.in not found, move up a directory and try again? This
455 could eliminate a common source of problems.
456 [ this is just a bad idea ]
458 * scripts are installed in $exec_prefix/bin, not $prefix/bin
460 [ the consensus on Gnits is that this isn't required.
461 doubters can work around it anyway ]
463 * make the auto-dep code crash if GNU make not in use?
464 (doesn't it already?)
466 Looked at a program called 'ezmake', which seems to do something
467 similar. The only idea there that is possibly worth stealing is using
468 globs in definitions. Also has negations. Eg in a directory with
469 files a.c, b.c and c.c, the line:
470 foo_SOURCES = *.c ~c.c
471 would be equivalent to:
472 foo_SOURCES = a.c b.c
473 Is this worth implementing?
474 [ No... it is more reliable to spell everything out. ]
476 Scan source directories and warn about missing files, eg .c/.h files
477 that aren't mentioned?
478 [ distcheck makes this less useful ]