1 Priorities for release:
2 !! documentation (eg new macros)
7 * copyrights on m4 files, aclocal output
9 * only remove libtool at top level?
11 * clean up source directory by moving stuff into subdirs
13 * consider adding pkglibexecdir, maybe others?
14 requests for pkg-dirs with version included
16 Fix ansi2knr so that knr-style function decls aren't incorrectly
17 rewritten. Or, fix automake so that not all sources in a directory
18 must be ansi. Should also fix things so that a separate ._o file is
19 not needed; instead use a wrapper script.
21 Gord on pretty-printing instead of raw loops:
22 Gord> I think the best thing to do is to have make output be a valid shell
23 Gord> script, and to have it be clear exactly what is happening at a given
29 - consider adding "echo"s to installs again.
30 users can use make SHELL='sh -x' to get the full dirt
32 Avoid loops when installing; instead unroll them in automake
34 Some long-term projects:
35 * if $(FOO) is used somewhere, ensure FOO is defined, either by
36 user or by automake if possible
37 * Don't rearrange order of `include' lines relative to += assignments.
38 * Handle += assignments at all.
40 consider putting all check-* targets onto @check?
41 To support --help/--version checking?
43 take diff-n-query code from libit
46 Per> 1) Being able to build a set of non-source programs
47 Per> from source programs, without necessarily linking them together.
48 Per> I.e. one should be able to say something like:
49 Per> dummy_SOURCES=foo.c bar.c
50 Per> and automake should realize that it needs to build foo.o and bar.o.
51 Per> 2) Being intelligent about new kinds of suffixes.
53 Per> SUFFIXES = .class .java
54 Per> and a suffix rule of the form:
56 Per> then it should be able to realize it can build .class files from
57 Per> .java files, and thus be able to generate a list of
58 Per> .class files from a list of .java source files.
61 * allow a way to use "gzip --best"
62 [ Try GZIP=--best make dist ]
63 * don't assume GNU tar is "tar" (eg in distcheck)
65 * actually use acinstall program
67 !! Must fix require_file stuff. It is really gross, and I don't
68 understand it any more.
70 * error messages should print ``[info blah blah]'' command when a
71 certain part of the standards apply. saw idea in message from
72 Craig Burley. wouldn't it be really cool if compile-mode in Emacs
73 understood this convention, and you could click on such text to
74 go to the appropriate info page?
76 !! should write autoconf-style doc entries for each m4 macro
78 Jim's idea: should look for @setfilename and warn if filenames too long
80 * allow ".info" to be missing
82 * must update GNU Hello
84 ** many requests for a way to omit a file from the distribution.
85 Should be done like `!foo' or `~foo' in _SOURCES, etc.
86 Such files should be removed explicitly after the copy step!
87 Doing this requires rewriting macros before generating Makefile.in.
89 from joerg-martin schwarz:
90 -- If Makefile.am contains $(CC), $(COMPILE), $(YLWRAP), ....
91 in an explicitly written rule, you should emit the corresponding
92 Makefile variables automatically.
94 Configuring in the large:
95 * add support for Makefile.tmpl that is auto-included in every
96 Makefile.am. That makes it easier to do some non-std thing in every
98 * allow hierarchy of dirs to share one aclocal.m4
101 consider printing full file name of Makefile.am or configure.in when
102 giving error. This would help for very large trees with many
105 From the GNU Standards. These things could be checked, and probably
107 * Make sure that the directory into which the distribution unpacks (as
108 well as any subdirectories) are all world-writable (octal mode 777).
109 * Make sure that no file name in the distribution is more than 14
111 * Don't include any symbolic links in the distribution itself.
113 * Make sure that all the files in the distribution are world-readable.
114 ** also, check --help output and --version output. Idea from François
115 * standards no longer prohibit ANSI C. What does this imply
116 for the de-ansi-fication feature?
118 consider supporting "var+= stuff" syntax. rewrite to just var=... on
119 output. This is sometimes convenient when you want to write a
120 Makefile.am in more-or-less modular parts
122 should be able to determine what is built by looking at rules (and
123 configure.in). Then built man pages (eg) could automatically be
124 omitted from the distribution.
126 Idea from Joerg-Martin Schwarz: allow passing different -D flags to
127 different compiles. This can be done, but with the restriction that a
128 .c cannot appear in 2 different "objects" (programs/libraries)
129 compiled with different -D options (because -c and -o do not always
130 work together and parallel makes must work). This could be
131 implemented by noticing whenever a ".o" target with no rules is being
132 emitted, and adding the appropriate compilation rule as appropriate.
133 This should work with targets from Makefile.am as well as from .P
134 files, which means rewriting so that the Makefile.am contents aren't
135 copied into the output immediately. This feature is probably required
136 to fully support libtool ("grody compilation issue")
137 [ this could be probably done more directly by examining the sources
138 as we scan Makefile.am ]
140 Henrik Frystyk Nielsen says:
141 Henrik> 4) Flags like --include-deps are lost when you make changes to
142 Henrik> Makefile.am files and automake is run automatically. It would
143 Henrik> be nice to keep these flags as I now have to redo everything
145 ... what about other options here too?
147 Think about: maybe "make check" should just bomb if error occurs?
148 Then user must use "make -k check". This is probably more natural.
150 Consider: "cvs" option adds some cvs-specific rules?
152 Right now, targets generated internally (eg "install") are not
153 overridable by user code. This should probably be possible, even
154 though it isn't very important. This could be done by generating all
155 internal rules via a function call instead of just appending to
157 [ this will be harder to implement when scanning a rule like all-recursive
160 * Should be a way to have "nobuild_PROGRAMS" which aren't even built,
161 but which could be by running the magic make command.
164 * Must rewrite am_install_var. Should break into multiple functions.
165 This will allow the callers to be a little smarter.
166 * Rewrite clean targets.
167 * Must rewrite error handling code. Right now it is a real mess
168 Should fix up require_file junk at the same time
170 Things to finish libtool support:
171 * Handle grody compilation issue
172 Namely: some objects require a different compilation rule
173 * Handle install changes
174 * Handle clean changes
175 * New definition for LINK
177 djm wants ``LINKS'' variable; list of things to link together after
178 install. In BSD environment, use:
179 LINKS = from1 to1 from2 to2 ...
181 Need way to say there are no suffixes in a Makefile (Franc,ois'
182 "override" idea suffices here)
184 Check to make sure various scripts are executable (IE when looking for
187 Use recode in dist target when MAINT_CHARSET specified. Read caveats
188 in automake.in before doing this. Note the same problem used to apply
189 to the no-dependencies option; maybe it still should? Note also that
190 each Makefile.am must be rewritten at "make dist" time if
191 MAINT_CHARSET and DIST_CHARSET are not identical. NOTE: gettext must
192 arrange for all .po files not to be recoded. In the long term this
193 might be a problem (consider when some systems use Unicode but the
195 MAINT_CHARSET *must* be local to each Makefile.am, to enable
196 merged distributions.
197 DIST_CHARSET must be passed down to subdir makes during a "make dist"
199 Handle dist-zoo. Generally add more DOS support. Maybe run "doschk"
200 (why isn't this merged with "pathchk"?) when doing a dist. Do
201 whatever else François says here...
203 Add support for html via an option. Use texi2html. Use
204 "html_TEXINFOS", and htmldir = .../html. Include html files in
205 distribution. Also allow "html_DATA", for raw .html files.
206 [ when will texinfo directly support html? ]
208 uninstall and pkg-dirs should rm -rf the dir.
210 a potential bug: configure puts "blah.o" into LIBOBJS, thus implying
211 these files can't be de-ansified. Not a problem?
212 [ fix by using ansi2knr wrapper program ]
214 In general most .am files should be merged into automake. For
215 instance all the "clean" targets could be merged by keeping lists of
216 things to be removed. This would be a lot nicer looking. Note that
217 the install targets probably should not be merged; it is sometimes
218 useful to only install a small part.
221 * Order rules sensibly
222 * Ensure every line has a purpose. Omit unused stuff
223 * Eliminate extraneous rules when possible (eg 'install-am' stuff)
224 * Make sure vertical spacing is correct
225 Omit program transform vars from header if no program installed. This
226 is currently pretty hard to do. (But with beautification code it
227 would probably be easy)
230 * It would be nice to automatically support using bison's better features
231 to rename the output files. This requires autoconf support
232 * Consider supporting syntax from autoconf "derived:source", eg:
234 for yacc and lex source
235 * allow generating c++ source from lex? Eg using flex?
237 Multi-language support:
238 * should have mapping of file extensions to languages
239 * should automatically handle the linking issue (special-case C++)
240 * must get compile rules for various languages; FORTRAN probably
241 most important unimplemented language
242 This should be integrated in some way with Per's idea.
243 Eg .f.o rules should be recognized & auto-handled in _SOURCES
244 That way any random language can be treated with C/C++ on a first-class
247 It might be cool to generate .texi dependencies by grepping for
248 @include. (If done, it should be done the same way C dependencies are
251 It would be good to check some parts of GNU standards. Already check
252 for install-sh and mkinstalldirs. What else is required to be in
253 package by GNU standards or by automake?
254 Some things for --strictness=gnits:
255 * "cd $(foo); something" is an error in a rule. Should be:
256 "cd $(foo) && something"
257 * Look for 'ln -s' and warn about using $(LN) and AC_PROG_LN_S
258 * Look for $(LN) and require AC_PROG_LN_S
260 Auto-distribute "ChangeLog.[0-9]+"? "ChangeLog.[a-z]+"?
262 Internationalize. [ gettext doesn't have the necessary machinery yet ]
263 am_error should use printf-style arguments (for eventual gettext scheme)
265 François says the ordering of files in a distribution should be as follows:
269 I agree, but I don't see how to implement this yet.
270 It might be easier if "derived files" is limited to those that
271 Automake itself knows about, eg output of yacc.
273 Check all source files to make sure that FSF address is up-to-date.
274 --gnits or --gnu only.
276 Merge each -vars.am file with corresponding ".am" file. Can do this
277 because of changes to &file_contents.
279 Should libexec programs have the name transform done on them?
281 Order the output rules sensibly, so FOO_SOURCES and FOO_OBJECTS are
282 together and rules are in the usual order.
284 Make the output minimal: only output definitions for variables that
287 Look at dist's jmake for ideas. dist is the name of the distribution
288 including Metaconfig. Perl uses it.
290 Should handle directory hierarchies deeper than 2. Right now there is
291 some support for this. Here are some of the issues:
292 * Should handle AC_CONFIG_SUBDIRS, ie must handle configure.in in subdirs
293 * can do this by looking at subdirs, seeing configure.in
294 and auto-running Automake there
296 . Consider supporting guile-style PLUGIN directories automatically?
299 David> To avoid comments like the one about subdirs getting buried in
300 David> the middle of a Makefile.in, how about pushing comments that
301 David> start with ### to the top of the Makefile.in (in order)? Sort
302 David> of like how Autoconf uses diversions to force initialization
303 David> code to the top of configure.
306 Karl> 2) Your Makefile variable names are generally uppercase, but GNU
307 Karl> generally uses lowercase. Not that it matters :-).
309 ================================================================
313 probably should put each group of m4 files into a subdir owned by the
314 containing application.
316 ================================================================
320 rationale for avoiding
321 make CFLAGS="$CFLAGS" ...
324 a package that installs its own aclocal macros
330 write example of using automake with dejagnu
331 follow calc example in dejagnu docs
333 document which variables are actually scanned and which are not.
335 Document customary ordering of Makefile.am. From François.
337 Should include extended version of diagram from Autoconf (suggested by
340 Make a definition of the term "source"
342 document how to use Automake with CVS. Idea from Mark Galassi. Also
343 include Greg Woods' more sophisticated "cvs-dist" target.
345 document rebuilding configure. CONFIGURE_DEPENDENCIES
347 -- must document all variables that are supposed
348 to be public knowledge
350 automake must be run in each directory with a configure.in
351 This is insufficiently clear
353 must document the targets required for integration with
354 non-automake-using subdirs
356 use of (eg) EXTRA_PROGRAMS is not very clear right now
358 document EXTRA_foo_SOURCES
360 document why EXTRA_* vars must be statically knowable
362 document the "make SHELL='/bin/sh -x'" trick for debugging
364 ================================================================
366 Things to do for autoconf:
368 * allow random code to be inserted a la the 2nd arg to AC_OUTPUT:
369 Joel> I know that the following is needed at the end of configure.in:
370 Joel> [test -z "$CONFIG_HEADERS" || echo timestamp >stamp-h])
371 Joel> However, if automake checked that this line is present, it would
372 Joel> help...this bit me for a while.
374 * patch autoreconf to run automake and aclocal (this is done but
375 not really available)
377 ================================================================
381 * Should support standalone library along with subdir library in same
382 Makefile.am. Maybe: turn off "standalone" mode if library's Makefile.am
383 is not only one specd? [ add an option for this ]
385 ================================================================
389 Would it be useful to integrate in some way with the Debian package
390 building utility? Must check. maybe it would be possible to deal
391 with all the different package utilities somehow. Lately I've been
392 hearing good things about the RedHat packaging utilities. Why are
393 there so many of these? Are they fun to write or something?
394 The RedHat package utility is called RPM; see
395 ftp://ftp.redhat.com/pub/code/rpm
396 It actually has problems, like no configure script and no documentation.
398 For Cygnus it would probably be good to be able to handle the native
399 package utility on each platform. There are probably 3 or 4 of these
400 (sysv, solaris?, aix?)
402 tcl/unix/Makefile.in has some code to generate a Solaris package.
404 Automake probably can't do all of this on its own. A new tool might
407 ================================================================
409 A tool to guess what the local Makefile.am should look like:
410 (see Gord's Maint program!)
412 * Probably integrate with autoscan
413 * Use various simple rules to determine what to do:
414 * get name of top directory, sans version info
415 * search for .c files with 'main' in them
416 * if in main.c, use directory name for program
417 * if in more than one, generate multiple programs
418 * if not found, generate a library named after directory
419 * order subdir searches correctly: lib first, src last
420 * assume 'testsuite' dir means we are using dejagnu
421 * maybe be smart about reading existing Makefile.am, so tool
422 can be run for incremental changes? You could imagine:
425 autoproject --incremental
427 ================================================================
429 Stuff NOT to do, and why:
431 consider auto-including any file that matches "*.in".
432 [ no: po/Makefile.in shouldn't be included ]
434 must look at mkid to see how it works (for subdir usage)
435 [ right now, it doesn't. i don't see a simple fix right now ]
437 if configure.in not found, move up a directory and try again? This
438 could eliminate a common source of problems.
439 [ this is just a bad idea ]
441 * scripts are installed in $exec_prefix/bin, not $prefix/bin
443 [ the consensus on Gnits is that this isn't required.
444 doubters can work around it anyway ]
446 * make the auto-dep code crash if GNU make not in use?
447 (doesn't it already?)
449 Looked at a program called 'ezmake', which seems to do something
450 similar. The only idea there that is possibly worth stealing is using
451 globs in definitions. Also has negations. Eg in a directory with
452 files a.c, b.c and c.c, the line:
453 foo_SOURCES = *.c ~c.c
454 would be equivalent to:
455 foo_SOURCES = a.c b.c
456 Is this worth implementing?
457 [ No... it is more reliable to spell everything out. ]
459 Scan source directories and warn about missing files, eg .c/.h files
460 that aren't mentioned?
461 [ distcheck makes this less useful ]