1 Priorities for release:
2 !! documentation (eg new macros)
3 * copyrights on m4 files, aclocal output
5 * should not put texiname_TEXINFOS into distribution
6 should rename this macro anyway, to foo_texi_DEPENDENCIES
8 For now I guess I'll just have automake give an error if it encounters
9 non-C source in a libtool library specification.
11 * must split $obj into two parts: one for libtool and one for
12 deansification. Otherwise .S files will be deansified!
14 ================================================================
21 * CONFIG_HEADER is passed incorrectly in ":" mode.
22 * put parser.h into distribution if "yacc -d" is used
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.
40 Avoid loops when installing; instead unroll them in automake
43 * completely handle multi-":" mode for AC_CONFIG_HEADER
44 * Scan multiple input files when Makefile is generated?
45 This would provide flexibility for large projects; subsumes
46 the "Makefile.tmpl" idea
48 [ can't do this. must explain why in manual.
49 basically, solving all the problems is too hard
50 like: how to remove redundancies between generated .in files
51 instead should implement `include' directive for Makefile.am ]
52 * for multi-":" mode and AC_OUTPUT, it might be good to pick the
53 first input file that has a corresponding .am file.
55 Some long-term projects:
56 * if $(FOO) is used somewhere, ensure FOO is defined, either by
57 user or by automake if possible
58 * Don't rearrange order of `include' lines relative to += assignments.
59 * Handle += assignments at all.
61 consider putting all check-* targets onto @check?
62 To support --help/--version checking?
64 take diff-n-query code from libit
67 Per> 1) Being able to build a set of non-source programs
68 Per> from source programs, without necessarily linking them together.
69 Per> I.e. one should be able to say something like:
70 Per> dummy_SOURCES=foo.c bar.c
71 Per> and automake should realize that it needs to build foo.o and bar.o.
72 Per> 2) Being intelligent about new kinds of suffixes.
74 Per> SUFFIXES = .class .java
75 Per> and a suffix rule of the form:
77 Per> then it should be able to realize it can build .class files from
78 Per> .java files, and thus be able to generate a list of
79 Per> .class files from a list of .java source files.
82 * allow a way to use "gzip --best"
83 [ Try GZIP=--best make dist ]
84 * don't assume GNU tar is "tar" (eg in distcheck)
86 !! Must fix require_file stuff. It is really gross, and I don't
87 understand it any more.
89 * error messages should print ``[info blah blah]'' command when a
90 certain part of the standards apply. saw idea in message from
91 Craig Burley. wouldn't it be really cool if compile-mode in Emacs
92 understood this convention, and you could click on such text to
93 go to the appropriate info page?
95 !! should write autoconf-style doc entries for each m4 macro
97 Jim's idea: should look for @setfilename and warn if filenames too long
99 * allow ".info" to be missing
101 * must update GNU Hello
103 ** many requests for a way to omit a file from the distribution.
104 Should be done like `!foo' or `~foo' in _SOURCES, etc.
105 Such files should be removed explicitly after the copy step!
106 Doing this requires rewriting macros before generating Makefile.in.
108 from joerg-martin schwarz:
109 -- If Makefile.am contains $(CC), $(COMPILE), $(YLWRAP), ....
110 in an explicitly written rule, you should emit the corresponding
111 Makefile variables automatically.
113 Configuring in the large:
114 * allow hierarchy of dirs to share one aclocal.m4
117 consider printing full file name of Makefile.am or configure.in when
118 giving error. This would help for very large trees with many
121 From the GNU Standards. These things could be checked, and probably
123 * Make sure that the directory into which the distribution unpacks (as
124 well as any subdirectories) are all world-writable (octal mode 777).
125 * Make sure that no file name in the distribution is more than 14
127 * Don't include any symbolic links in the distribution itself.
129 * Make sure that all the files in the distribution are world-readable.
130 ** also, check --help output and --version output. Idea from François
131 * standards no longer prohibit ANSI C. What does this imply
132 for the de-ansi-fication feature?
134 consider supporting "var+= stuff" syntax. rewrite to just var=... on
135 output. This is sometimes convenient when you want to write a
136 Makefile.am in more-or-less modular parts
138 should be able to determine what is built by looking at rules (and
139 configure.in). Then built man pages (eg) could automatically be
140 omitted from the distribution.
142 Idea from Joerg-Martin Schwarz: allow passing different -D flags to
143 different compiles. This can be done, but with the restriction that a
144 .c cannot appear in 2 different "objects" (programs/libraries)
145 compiled with different -D options (because -c and -o do not always
146 work together and parallel makes must work). This could be
147 implemented by noticing whenever a ".o" target with no rules is being
148 emitted, and adding the appropriate compilation rule as appropriate.
149 This should work with targets from Makefile.am as well as from .P
150 files, which means rewriting so that the Makefile.am contents aren't
151 copied into the output immediately.
152 [ this could be probably done more directly by examining the sources
153 as we scan Makefile.am ]
155 Henrik Frystyk Nielsen says:
156 Henrik> 4) Flags like --include-deps are lost when you make changes to
157 Henrik> Makefile.am files and automake is run automatically. It would
158 Henrik> be nice to keep these flags as I now have to redo everything
160 ... what about other options here too?
162 Think about: maybe "make check" should just bomb if error occurs?
163 Then user must use "make -k check". This is probably more natural.
165 Consider: "cvs" option adds some cvs-specific rules?
167 Right now, targets generated internally (eg "install") are not
168 overridable by user code. This should probably be possible, even
169 though it isn't very important. This could be done by generating all
170 internal rules via a function call instead of just appending to
172 [ this will be harder to implement when scanning a rule like all-recursive
175 * Should be a way to have "nobuild_PROGRAMS" which aren't even built,
176 but which could be by running the magic make command.
179 * Must rewrite am_install_var. Should break into multiple functions.
180 This will allow the callers to be a little smarter.
181 * Rewrite clean targets.
182 * Must rewrite error handling code. Right now it is a real mess
183 Should fix up require_file junk at the same time
185 djm wants ``LINKS'' variable; list of things to link together after
186 install. In BSD environment, use:
187 LINKS = from1 to1 from2 to2 ...
189 Need way to say there are no suffixes in a Makefile (Franc,ois'
190 "override" idea suffices here)
192 Check to make sure various scripts are executable (IE when looking for
195 Use recode in dist target when MAINT_CHARSET specified. Read caveats
196 in automake.in before doing this. Note the same problem used to apply
197 to the no-dependencies option; maybe it still should? Note also that
198 each Makefile.am must be rewritten at "make dist" time if
199 MAINT_CHARSET and DIST_CHARSET are not identical. NOTE: gettext must
200 arrange for all .po files not to be recoded. In the long term this
201 might be a problem (consider when some systems use Unicode but the
203 MAINT_CHARSET *must* be local to each Makefile.am, to enable
204 merged distributions.
205 DIST_CHARSET must be passed down to subdir makes during a "make dist"
207 Handle dist-zoo. Generally add more DOS support. Maybe run "doschk"
208 (why isn't this merged with "pathchk"?) when doing a dist. Do
209 whatever else François says here...
211 Add support for html via an option. Use texi2html. Use
212 "html_TEXINFOS", and htmldir = .../html. Include html files in
213 distribution. Also allow "html_DATA", for raw .html files.
214 [ when will texinfo directly support html? ]
216 uninstall and pkg-dirs should rm -rf the dir.
218 a potential bug: configure puts "blah.o" into LIBOBJS, thus implying
219 these files can't be de-ansified. Not a problem?
220 [ fix by using ansi2knr wrapper program ]
222 In general most .am files should be merged into automake. For
223 instance all the "clean" targets could be merged by keeping lists of
224 things to be removed. This would be a lot nicer looking. Note that
225 the install targets probably should not be merged; it is sometimes
226 useful to only install a small part.
229 * Order rules sensibly
230 * Ensure every line has a purpose. Omit unused stuff
231 * Eliminate extraneous rules when possible (eg 'install-am' stuff)
232 * Make sure vertical spacing is correct
233 Omit program transform vars from header if no program installed. This
234 is currently pretty hard to do. (But with beautification code it
235 would probably be easy)
238 * It would be nice to automatically support using bison's better features
239 to rename the output files. This requires autoconf support
240 * Consider supporting syntax from autoconf "derived:source", eg:
242 for yacc and lex source
243 * allow generating c++ source from lex? Eg using flex?
245 Multi-language support:
246 * should have mapping of file extensions to languages
247 * should automatically handle the linking issue (special-case C++)
248 * must get compile rules for various languages; FORTRAN probably
249 most important unimplemented language
250 This should be integrated in some way with Per's idea.
251 Eg .f.o rules should be recognized & auto-handled in _SOURCES
252 That way any random language can be treated with C/C++ on a first-class
255 It might be cool to generate .texi dependencies by grepping for
256 @include. (If done, it should be done the same way C dependencies are
259 It would be good to check some parts of GNU standards. Already check
260 for install-sh and mkinstalldirs. What else is required to be in
261 package by GNU standards or by automake?
262 Some things for --strictness=gnits:
263 * "cd $(foo); something" is an error in a rule. Should be:
264 "cd $(foo) && something"
265 * Look for 'ln -s' and warn about using $(LN) and AC_PROG_LN_S
266 * Look for $(LN) and require AC_PROG_LN_S
268 Auto-distribute "ChangeLog.[0-9]+"? "ChangeLog.[a-z]+"?
270 Internationalize. [ gettext doesn't have the necessary machinery yet ]
271 am_error should use printf-style arguments (for eventual gettext scheme)
273 François says the ordering of files in a distribution should be as follows:
277 I agree, but I don't see how to implement this yet.
278 It might be easier if "derived files" is limited to those that
279 Automake itself knows about, eg output of yacc.
281 Check all source files to make sure that FSF address is up-to-date.
282 --gnits or --gnu only.
284 Merge each -vars.am file with corresponding ".am" file. Can do this
285 because of changes to &file_contents.
287 Should libexec programs have the name transform done on them?
289 Order the output rules sensibly, so FOO_SOURCES and FOO_OBJECTS are
290 together and rules are in the usual order.
292 Make the output minimal: only output definitions for variables that
295 Look at dist's jmake for ideas. dist is the name of the distribution
296 including Metaconfig. Perl uses it.
298 Should handle directory hierarchies deeper than 2. Right now there is
299 some support for this. Here are some of the issues:
300 * Should handle AC_CONFIG_SUBDIRS, ie must handle configure.in in subdirs
301 * can do this by looking at subdirs, seeing configure.in
302 and auto-running Automake there
305 David> To avoid comments like the one about subdirs getting buried in
306 David> the middle of a Makefile.in, how about pushing comments that
307 David> start with ### to the top of the Makefile.in (in order)? Sort
308 David> of like how Autoconf uses diversions to force initialization
309 David> code to the top of configure.
312 Karl> 2) Your Makefile variable names are generally uppercase, but GNU
313 Karl> generally uses lowercase. Not that it matters :-).
315 ================================================================
319 probably should put each group of m4 files into a subdir owned by the
320 containing application.
322 ================================================================
326 multi-":" mode in AC_OUTPUT -- automake only looks at the first file
327 also a note on how a .am file is found in this case
329 rationale for avoiding
330 make CFLAGS="$CFLAGS" ...
333 a package that installs its own aclocal macros
335 write example of using automake with dejagnu
336 follow calc example in dejagnu docs
338 document which variables are actually scanned and which are not.
340 Document customary ordering of Makefile.am. From François.
342 Should include extended version of diagram from Autoconf (suggested by
345 Make a definition of the term "source"
347 document how to use Automake with CVS. Idea from Mark Galassi. Also
348 include Greg Woods' more sophisticated "cvs-dist" target.
350 document rebuilding configure. CONFIGURE_DEPENDENCIES
352 -- must document all variables that are supposed
353 to be public knowledge
355 automake must be run in each directory with a configure.in
356 This is insufficiently clear
358 must document the targets required for integration with
359 non-automake-using subdirs
361 use of (eg) EXTRA_PROGRAMS is not very clear right now
362 document EXTRA_foo_SOURCES
363 document why EXTRA_* vars must be statically knowable
365 document the "make SHELL='/bin/sh -x'" trick for debugging
367 section on relationship to GNU make
371 ================================================================
373 Things to do for autoconf:
375 * patch autoreconf to run automake and aclocal. I've done this but it is
376 not really available. It can't be made available until automake
377 is officially released
379 ================================================================
383 * Should support standalone library along with subdir library in same
384 Makefile.am. Maybe: turn off "standalone" mode if library's Makefile.am
385 is not only one specd? [ add an option for this ]
387 ================================================================
391 Would it be useful to integrate in some way with the Debian package
392 building utility? Must check. maybe it would be possible to deal
393 with all the different package utilities somehow. Lately I've been
394 hearing good things about the RedHat packaging utilities. Why are
395 there so many of these? Are they fun to write or something?
396 The RedHat package utility is called RPM; see
397 ftp://ftp.redhat.com/pub/code/rpm
398 It actually has problems, like no configure script and no documentation.
400 For Cygnus it would probably be good to be able to handle the native
401 package utility on each platform. There are probably 3 or 4 of these
402 (sysv, solaris?, aix?)
404 tcl/unix/Makefile.in has some code to generate a Solaris package.
406 Automake probably can't do all of this on its own. A new tool might
409 ================================================================
411 A tool to guess what the local Makefile.am should look like:
412 (see Gord's Maint program!)
414 * Probably integrate with autoscan
415 * Use various simple rules to determine what to do:
416 * get name of top directory, sans version info
417 * search for .c files with 'main' in them
418 * if in main.c, use directory name for program
419 * if in more than one, generate multiple programs
420 * if not found, generate a library named after directory
421 * order subdir searches correctly: lib first, src last
422 * assume 'testsuite' dir means we are using dejagnu
423 * maybe be smart about reading existing Makefile.am, so tool
424 can be run for incremental changes? You could imagine:
427 autoproject --incremental
429 ================================================================
431 Stuff NOT to do, and why:
433 consider auto-including any file that matches "*.in".
434 [ no: po/Makefile.in shouldn't be included ]
436 must look at mkid to see how it works (for subdir usage)
437 [ right now, it doesn't. i don't see a simple fix right now ]
439 if configure.in not found, move up a directory and try again? This
440 could eliminate a common source of problems.
441 [ this is just a bad idea ]
443 * scripts are installed in $exec_prefix/bin, not $prefix/bin
445 [ the consensus on Gnits is that this isn't required.
446 doubters can work around it anyway ]
448 * make the auto-dep code crash if GNU make not in use?
449 (doesn't it already?)
451 Looked at a program called 'ezmake', which seems to do something
452 similar. The only idea there that is possibly worth stealing is using
453 globs in definitions. Also has negations. Eg in a directory with
454 files a.c, b.c and c.c, the line:
455 foo_SOURCES = *.c ~c.c
456 would be equivalent to:
457 foo_SOURCES = a.c b.c
458 Is this worth implementing?
459 [ No... it is more reliable to spell everything out. ]
461 Scan source directories and warn about missing files, eg .c/.h files
462 that aren't mentioned?
463 [ distcheck makes this less useful ]