"" == Shigio Yamaguchi [EMAIL PROTECTED] writes:
In BSD system, '/usr/obj' + current directory is used for build
directory. (from now on, I say it 'obj' directory.)
If there is a convention in automake, I can support it.
There isn't a simple convention like this.
The person building the
Derek What's $(top_builddir)/.deps get used for when there aren't any C
Derek sources in $(top_srcdir) or $(top_builddir)?
Thanks, I checked in the fix.
Tom
Derek Whoops. Forgot to put [PATCH] in the subject line.
Don't worry too much about that convention.
Anyway, I already checked in a slightly different fix for the bug.
Tom
"Derek" == Derek R Price [EMAIL PROTECTED] writes:
Derek I added rudimentary support for different implementations of
Derek etags (read the one Automake expects and Exuberent etags) since
Derek they take slightly different options. Exuberent etags is the
Derek version distributed with RedHat
"Derek" == Derek R Price [EMAIL PROTECTED] writes:
Derek Is there a good reason the configure script creates
Derek $(top_builddir)/.deps during the test that sets $DEPDIR and
Derek doesn't delete it again? Besides some developer or other
Derek needing sleep? ;)
Derek My distribution
"Kevin" == Kevin Ryde [EMAIL PROTECTED] writes:
Kevin I'd like to propose the patch below to put any $(SUFFIXES) from
Kevin Makefile.am at the head of the .SUFFIXES list in Makefile.in.
Thanks, I finally checked this in.
Tom
"Patrick" == Patrick Guio [EMAIL PROTECTED] writes:
Patrick I am not sure if this is to be adressed to libtool or maybe
Patrick automake since the problem might come from depcom... Anyway
Patrick I am trying to build a c++ library using libtool on a
Patrick alphaev6-dec-osf5.0 using cxx.
Did
"Braden" == Braden McDaniel [EMAIL PROTECTED] writes:
Tom I'd really like to find a better solution for the built files
Tom problem. This is currently too hard to do in automake.
Braden I ended up doing this:
Braden $(libfoo_la_OBJECTS): mybuiltheader.h
I recently changed how BUILT_SOURCES
"Tim" == Tim Heath [EMAIL PROTECTED] writes:
Tim CXX=aCC CC=aCC ./configure --enable-dependency-tracking
Tim Is this the correct way to use thise new dependency generation?
Yes.
Tim When I touch a .C that should cause other .o's to be regenerated
Tim it is not occurring. It does not attempt
"Walter" == Walter Lievens [EMAIL PROTECTED] writes:
Walter Since I want to install MySQL++ (the C++ API of MySQL), I need
Walter to use automake.
I'm a bit suprised. Ordinarily a package maintainer uses `make dist';
then the person building the package doesn't need automake.
Walter
"Derek" == Derek R Price [EMAIL PROTECTED] writes:
Me too. But the point is that GNU packages are supposed to ship
with texinfo.tex.
Derek Is there a web page somewhere with this standard on it? I
Derek browsed briefly but I haven't been able to locate one.
It is in the GNU Coding
"Bryan" == Bryan VanDeVen [EMAIL PROTECTED] writes:
Bryan what is the best way to use a different source file based on
Bryan the outcome of configure? We would like to normailly produce a
Bryan stubbed out version of a lib function, and only use the real
Bryan version when building inside the
"John" == John R Cary [EMAIL PROTECTED] writes:
John Are there any plans for a new release of automake?
We wish.
John There is an old problem, mentioned at
John http://sources.redhat.com/ml/automake/1999-05/msg00011.html with
John the choice of AR for C++ libraries.
The CVS automake has a
Pavel I have checked (manually) all Automake *.m4 files for
Pavel underquoting. AM_PROG_REGEX was actually broken. Now it's
Pavel fixed.
Thanks.
Tom
"Benoit" == benoit LOOF [EMAIL PROTECTED] writes:
Benoit Does anybody know when automake 1.5 will be available ?
Nobody knows.
Tom
"Tim" == Tim Heath [EMAIL PROTECTED] writes:
TimI execute the command with the following output:
Tim cvs -z 9 -d :pserver:[EMAIL PROTECTED]:/cvs/automake login
Tim (Logging in to [EMAIL PROTECTED])
Tim CVS password:
Tim cvs [login aborted]: connect to anoncvs.cygnus.com:2401 failed:
Tim
"Marc" == Marc van Woerkom [EMAIL PROTECTED] writes:
Marc MY_MODULES="Base Dataserver2"
Marc MY_MAKEFILES="`for D in $MY_MODULES; do echo $ac_n ' '$D/Makefile; done`"
Marc AC_OUTPUT([
Marc Makefile
Marc macros/Makefile
Marc $MY_MAKEFILES
Marc src/Makefile
Marc
"Braden" == Braden McDaniel [EMAIL PROTECTED] writes:
I realize that BUILT_SOURCES is known to be problematic, but I haven't
found any other solution. Any suggestions?
Add explicit dependencies on it. There's no other way to ensure it's
generated. BUILT_SOURCES used to work only with
"Fred" == F Labrosse [EMAIL PROTECTED] writes:
Fred I found a way to do all that. My only whish, for now, is that I
Fred would like to have a new kind of target: OBJECTS to only build
Fred .o files and not complete programs or libraries. Can you
Fred comment on that?
There have been a couple
"Marc" == User Marc van Woerkom [EMAIL PROTECTED] writes:
Marc I have a question regarding automake 1.4/1.4a:
Marc Is it possible to build a single library from source files
Marc that reside in a couple of subdirectories beneath?
This is possible with the cvs automake.
Tom
"Paul" == Paul Martinolich [EMAIL PROTECTED] writes:
Paul Using the CVS version of automake, I am getting spurious output
Paul from the dependency checking /usr/local/share/aclocal/depend.m4
Paul line 50.
Thanks for the report and the patch. I'm going to check it in.
Tom
"Pavel" == Pavel Roskin [EMAIL PROTECTED] writes:
check_PROGRAMS = test_program
test_program_SOURCES = test.C
Pavel You have discovered a bug in Automake. It still exists in the
Pavel current CVS version.
Pavel Don't start lines in Makefile.am with spaces. Automake will not
Pavel
"Melissa" == Brown, Melissa [EMAIL PROTECTED] writes:
Melissa Here's the error I get when doing the make check on
Melissa automake-1.4:
Melissa .
Melissa .
Melissa .
Melissa PASS: yaccpp.test
Melissa =
Melissa 2 of 194 tests failed
Melissa =
Look at the
"Simon" == Simon Richter [EMAIL PROTECTED]
writes:
On Fri, 17 Nov 2000, Tim Heath wrote:
How do I add compiler flags? What AM is necessary to accomplish that?
Simon Make your configure.in add them to $CFLAGS, before AC_PROG_CC
Simon (to make sure the compiler actually accepts them).
"Benoit" == benoit LOOF [EMAIL PROTECTED] writes:
Benoit I'd lihe to know if there any gurther documentation about automake.
Benoit I've looked at www.gnu.org/software/automake. Is there any other
Benoit location ?
Last time I looked, this site is out of date. This is the source of a
Pavel
Pavel If you omit the space before the path, then 4.2BSD based systems
Pavel (such as Sequent DYNIX) will ignore the line, because they interpret
Pavel `#! /' as a 4-byte magic number.
Pavel
Reliable sources tell me that this is a myth albeit a
"Allen" == [EMAIL PROTECTED] writes:
Allen I would like to be able to build both debug and release builds
Allen of some libraries and apps that are using the GNU build
Allen system. It seems that there must be an easy way to specify
Allen multiple targets (one simply stripped of symbols). Can
"Pavel" == Pavel Roskin [EMAIL PROTECTED] writes:
Pavel :pserver:[EMAIL PROTECTED]:/cvs/automake
This is the master repository.
Pavel :pserver:[EMAIL PROTECTED]:/cvs
They mirror the repo here.
Pavel This is getting confusing. One of the repositories should be
Pavel killed. Login on that
"john" == john heasley [EMAIL PROTECTED] writes:
john is there a macro or option to automake to ommit the target
john for $(srcdir)/Makefile.in, such that it is not necessary to
john include Makefile.am and configure.in in a distribution?
FYI, we don't support not distributing these files. It
"Clark" == Clark Rawlins [EMAIL PROTECTED] writes:
Clark When I went to generate my Makefile using configure I expected
Clark that my libdir variable would be set to ${prefix}/lib not
Clark $(shell pwd). What is going on? What other information is
Clark relavant? Is this the right place to
"James" == James Bonfield [EMAIL PROTECTED] writes:
James eg_SOURCES = a/a.c b/b.c c/c.c
As Alexandre says, this works in the cvs automake.
James Notice the lack of a plus sign.
Automake handles `+=' "specially", which unfortunately means it
sometimes doesn't do what you expect.
James In
"Alex" == Alex Hornby [EMAIL PROTECTED] writes:
Alex Here is the first broken out part of my parallel built sources
Alex patch, which I agreed to split up and resubmit back in the midst
Alex of time. I hope you still have the time to review these.
It is time for us to start writing test code
"Tim" == Tim Heath [EMAIL PROTECTED] writes:
Tim It appears that when I modify Makefile.am it will then
Tim automatically modify Makefile.in. If I modify Makefile.in it
Tim will automatically cause the Makefile to be regenerated. How can
Tim I stop that?
Generally speaking, you don't really
Pavel ldflags.test is a new test and should probably be added to
Pavel XFAIL_TESTS.
Done.
Pavel lex4.test is broken by the last change committed by Tom Tromey
Pavel on 2000-08-27. For some reason joe.Po is no longer present in
Pavel Makefile.in
Fixed. Thanks.
Tom
Right now `make clean' cleans directories in the reverse of the order
it builds them. This is mostly correct. However, there are a couple
of automake PRs that point out that for subdir cleans it would be
better to clean depth first. Right now we clean `.' before the
subdirs by default.
I'm
"Kevin" == Kevin Ryde [EMAIL PROTECTED] writes:
Kevin Using a recent cvs automake, I tried "make TAGS" in a separate
Kevin object directory, but the etags command got config.in without a
Kevin $(srcdir) path. Perhaps the @CONFIG@ in tags.am should be in
Kevin the list that gets uniquified and
"" == Jiann-Ming Su [EMAIL PROTECTED] writes:
Is automake's dependency tracking smart enough to know when a
header file has been removed? I'm currently using 1.4a.
There is no official 1.4a. You're probably using Red Hat.
In 1.4 automake was not smart enough to detect this situation.
This
Murray However, if the share/aclocal/something.m4 is installed in
Murray under a --prefix which is not where aclocal is installed, how
Murray should I (or the user) tell aclocal to look in this second
Murray directory as well as it's normal share/aclocal/ directory?
aclocal -I /the/directory
Peter Automake-generated makefiles should somehow be coping with the
Peter possibility that install-info is not available. Currently, the
Peter uninstall target of CVS Autoconf is doesn't work for me because
Peter of this. (The `info' RPM somehow placed it at
Peter /sbin/install-info.)
Can you
Martin And then, for full functionality, automake should use that
Martin variable in generating rules, right?
Martin .c.o:
Martin $(CC) $(CFLAGS) ... @CC_C_O@ $
This won't be sufficient.
Automake will need more information.
Tom
Martin Wrt to supporting every compiler, let's forget about it. So
Martin far I haven't seen a single compiler with decent dependency
Martin tracking support. And even if they did, you can be certain
Martin that each and every one would use a different option syntax
Martin and output format.
We
"Tim" == Tim Heath [EMAIL PROTECTED] writes:
Tim I am interested in combining libraries created in a deep project
Tim into one library instead of a bunch of little ones. How can this
Tim be accomplished?
Use libtool convenience libraries for the subdirectories, and combine
them into a single
Martin I understand you're trying to get a new release out.
It isn't that. It's just that I (and, apparently, the other automake
hackers) have very little time.
Tom
Jerome I try to write a transportable AC_DEFUN:
Jerome I want to wrire simple substitution command as (in bash):
Jerome $bb=${BB%%.*}
Jerome Is there a more transportable way to do that ?
This is more of an autoconf question.
Generally the answer is "use sed".
Do you really mean to assign to
I removed the autoconf list from the followups. Dependency tracking
doesn't involve autoconf.
Martin First of all, an automake-generated makefile refuses to work
Martin if there is no C source file specified in Makefile.am (only
Martin Fortran source files).
Could you submit a bug report to
Alain It was/is my understanding that automake honoured DESTDIR.
It does.
Any failure of DESTDIR is most likely a bug in a particular package's
Makefile.am. If not, then it is an automake bug and should be
reported.
Tom
Andy I am trying to use automake 1.4 on an existing software package
Andy whose source directory are organized as a set of subdirectories.
Andy How can I write a makefile.am to automatically generate a
Andy library file from these source files?
The easiest way in 1.4 is to make a libtool
Martin If I understand the automake philosophy right, then it
Martin suffices to support only a few compilers, generate the
Martin dependencies on the maintainer's site, and include them into
Martin the distributed Makefiles.
Actually, ideally we'd like to support every compiler. However, I
Jerome Where can we get a large variety of `Makefile.am' scripts ?
Download a lot of GNU packages.
Tom
Murray Apparently I should do with this with a noinst_LTLIBRARIES, so
Murray you do need to correct the bit in the automake manual that
Murray says that it isn't allowed.
I think that has already been fixed.
When I looked at my copy of the manual this morning it said that
check_LTLIBRARIES is
Mo Paul, could you post a link to your project? I am going to be
Mo rewriting the build system for the tcljava project using automake,
Mo so I would also like to take a peek at some examples. I already
Mo have macros to do the JVM detection, I am just interested in the
Mo .java and .jar rules you
Peter Alternatively, the GNU standards could catch up with reality
Peter and make `docdir' a standard directory. Not everyone uses
Peter Texinfo.
For GNU Standards changes you'd have to write to RMS.
Tom
Lars What I want to know if the config.status problem was the only
Lars reason why the SUBDIR variable should only contain single-level
Lars directory names? In that case, the "directory should not
Lars contain '/'" warning could be removed. If there are more
Lars problems, what are they?
I
Alexandre In particular, in the case of installing the library, you
Alexandre must add the -rpath switch to LDFLAGS_LIBJS. I believe
Alexandre that's the main reason why automake doesn't support
Alexandre conditional installation of ltlibraries.
Yes, that's right.
Is there something we can do
"Jonas" == Jonas Berlin [EMAIL PROTECTED] writes:
Jonas I don't know if this has been fixed or not, but in my
Jonas automake-1.4-5 which came with RedHat 6.1, there are bugs in
Jonas the main executable, i.e. "automake".
Both the bugs you mention are fixed in the cvs automake.
Jonas And then
Ossama There may be an easier to achieve what you want. How about
Ossama the following:
Ossama pkgincludedir = $(prefix)/include/PackageName
This is wrong in two ways.
Use $(includedir)/PackageName instead.
Tom
Ossama pkgincludedir = $(prefix)/include/PackageName
This is wrong in two ways.
Use $(includedir)/PackageName instead.
Ossama Thanks for clarification! That makes more sense! What's the
Ossama second way it is wrong?
Ossama pkgincludedir = $(includedir)/PackageName
Ossama
"Alex" == Alex Hornby [EMAIL PROTECTED] writes:
Alex Is there already an expand_make_variable() type function?
Yes. You can use variable_value_as_list in automake to do this. But
really you'll want to have direct support for IDL in automake, so that
IDL files can be listed directly in
"Alex" == Alex Hornby [EMAIL PROTECTED] writes:
Alex As IDL files use standard C preprocessor syntax for dependent
Alex inclusion, is it not possible to wrap the idl compile in
Alex depcomp?
Interesting idea. I thought different IDL compilers generated
different output file names, though. Is
.
I think this problem is fixed in cvs. If it is the problem I am
thinking of, then it is actually a combination of a BSD make oddity
(it runs "sh -e") and a BSD sh oddity (it treats and || strangely).
I think the problem was fixed by this change:
1999-03-02 Tom Trom
1. I can't get automake to read a dynamically created AC_OUTPUT
parameter list. This is correct behavior but how do I specify
which files it should create?
Yes, automake needs a static AC_OUTPUT list. Automake uses the
argument to AC_OUTPUT to determine which Makefile.am's to scan.
"John" == John R Sheets [EMAIL PROTECTED] writes:
John Is it possible to change the VERSION #define to something different?
Use the third argument to AM_INIT_AUTOMAKE to disable the define.
Then AC_DEFINE anything else you want.
Tom
John No, I haven't tried CVS automake, but that's not really an
John option for this project. It'll be an uphill battle to get my
John automake-ifying patches accepted in the first place (to the WINE
John project), so I'm pretty much stuck with released versions.
I totally understand. It would
"Bram" == Bram Stolk [EMAIL PROTECTED] writes:
Bram My own autoconf 2.13 generated configure script fails, due to
Bram unknown --build option to ltconfig (not present in Mesa's
Bram configure).
You probably have a mismatch between your libtool m4 macro and your
ltconfig version. I'd ask on
"Akim" == Akim Demaille [EMAIL PROTECTED] writes:
Akim Honestly, I see no difference between plenty of small utilities
Akim and one big. The difference is moot to me (Hm, I read that
Akim somewhere, but I'm not sure I'm writing correct English :).
There's a difference to me. I get to hear
"David" == Masterson, David [EMAIL PROTECTED] writes:
David Would those complaints be really due to "too many files"
David (ie. size issue) or "too many files with questionable purpose"?
Actually I think it is not size (in bytes) or purpose, but rather the
annoyance of typing "ls" and seeing
"Peter" == Peter Eisentraut [EMAIL PROTECTED] writes:
I think we should not introduce more utilities. We have way too
many already. How about reimplementing this as a patch to
"missing"?
Peter Hmm, "missing" _fakes_ a given program (with touch, etc.)
Peter whereas mkinstalldirs,
"Peter" == Peter Eisentraut [EMAIL PROTECTED] writes:
Peter Might part of that perceived uselessness come from the fact
Peter that it doesn't remove the directories?
I don't think so. I think that one big reason is that you have to
keep (the skeleton of) the build tree around for "uninstall"
"Peter" == Peter Eisentraut [EMAIL PROTECTED] writes:
Peter The added beauty of this approach would be that `rminstalldirs'
Peter would be to `rmdir -p' what `mkinstalldirs' is now to `mkdir
Peter -p'. A possible (though perhaps not perfect) implementation of
Peter rminstalldirs is attached for
"Pavel" == Pavel Roskin [EMAIL PROTECTED] writes:
Pavel Let's make "missing" handle "mkdir -p" to begin with.
That would be fine with me.
Pavel I still insist that no directories should be removed by "make
Pavel uninstall" unless they defininely belong to the package being
Pavel uninstalled.
Michael Does anyone have an example that uses automake's multilib
Michael feature that I could look at?
The automake multilib support isn't really complete.
Tom
"David" == Masterson, David [EMAIL PROTECTED] writes:
David If you want, I have a Perl tool that I picked up and enhanced
David called PKGLINK that could help with this. Basically, with
David pkglink, you install all packages in there own prefix directory
David and then use pkglink to
JMarc Thanks for the tip. Does it mean that what I am trying to do is
JMarc not a good idea, or that automake is lacking a feature?
I'm not sure it is either of these.
JMarc What I mean is that I am ready to provide a more "standard" way
JMarc of installing several LyX versions concurrently,
"JMarc" == Jean-Marc Lasgouttes [EMAIL PROTECTED] writes:
JMarc With autoconf, since I want to use the autogenerated makefile
JMarc rules, I figured out that setting PACKAGE to the right value at
JMarc configure time should work. Unfortunately, I cannot get the
JMarc name of the installed
"David" == Masterson, David [EMAIL PROTECTED] writes:
David * Does Automake assume that Info files should be part of the
David distribution of a package (ie. "make dist")?
Yes.
David ** If so, why?
GNU coding standards
David ** Is there a means to specify that Info files should *not* be
"Axel" == Axel Boness [EMAIL PROTECTED] writes:
Axel I tried things like :
Axel MYDATA=@prefix@/@VERSION@/@PACKAGE@
Axel AC_SUBST(MYDATA)
This is definitely in the archives. It comes up all the time.
One answer is that you should use a sed rule in Makefile, and not
AC_OUTPUT, if you want full
"Pavel" == Pavel Roskin [EMAIL PROTECTED] writes:
Pavel The attached patch fixes the testsuite. Currently some tests
Pavel use some files (depcomp, install-sh, mkinstalldirs and missing)
Pavel from the toplevel Automake directory if and only if srcdir=="."
Thanks, I checked this in.
Pavel I
"dv" == Didier Verna [EMAIL PROTECTED] writes:
dv The latest cvs'ed automake doesn't appear to support autoconf's
dv AC_CONFIG_FILES. I've seen an old thread about this (jan 2000) in
dv the archives, but is there a patch in the queue for this ?
There is. I thought Jim had checked it in.
Tom
"Ted" == Ted Irons [EMAIL PROTECTED] writes:
Ted Does the multi-language-branch of libtool
Ted work with autoconf-2.13 and automake-1.4?
I haven't tried it, but I don't see why it wouldn't.
I think Alexandre might do this already.
Give it a try and tell us what you find out...
T
"Braden" == Braden N McDaniel [EMAIL PROTECTED] writes:
Braden I have hacked around this problem by emptying the MY_HEADERS
Braden variable explicitly in every Makefile.am in the project except
Braden for the one where it's actually pertinent, but I have the
Braden feeling that either this is a
Alex Please disregard my earlier patch. I think all that needs adding
Alex is a warning when there are C++ subdir sources without
Alex subdir-objects being set.
This case is supposed to work as well, though.
Probably rule replication is the way to go.
I can't look at this in the near future.
"Lars" == Lars J Aas [EMAIL PROTECTED] writes:
Lars I've always been annoyed that automake --add-missing --copy
Lars doesn't pass the "--copy"-option along to libtoolize, so I end
Lars up with symlinked config.guess, config.sub, ltconfig and
Lars ltmain.sh.
Fixed.
Lars This might be fixed
Sam The basic problem is that source is conditionally compiled into
Sam SDL, and supposedly the attached code is supposed to do the
Sam "right thing", but I can't get it to include rules for the
Sam objects. I can get the desired effect using automake
Sam conditionals, but the complexity of
"Sascha" == Sascha Ziemann [EMAIL PROTECTED] writes:
Sascha flex conf_lexer.l mv lex.yy.c conf_lexer.c
Sascha gcc -c conf_lexer.c
Sascha bison -y conf_parser.y mv y.tab.c conf_parser.c
Sascha gcc -c conf_parser.c
Sascha And this means, that the tokens are not available in the
Sascha
"Alex" == Alex Hornby [EMAIL PROTECTED] writes:
How about just creating standard Makefile dependencies?
Foo_impl.o: Foo_s.hh Foo_c.hh Foo_s.cpp Foo_c.cpp
Foo_impl.$(OBJEXT): Foo_s.hh Foo_c.hh Foo_s.cpp Foo_c.cpp
Alex That would work. I wanted to avoid adding dependencies manually
Alex
"" == OKUJI Yoshinori [EMAIL PROTECTED] writes:
2. The target `Makefile' depends on $(BUILT_SOURCES).
The solution may be to get rid of the dependency in the item 2,
so my question is why Makefile must depend on BUILT_SOURCES. Is
there some good reason?
Only a historical one. In the
"Olly" == Olly Betts [EMAIL PROTECTED] writes:
localhost:~/src/4dim$ automake
Makefile.am:2: bad macro name '4dim_SOURCES'
Olly I hit this problem a few months ago. It's an unnecessary
Olly restriction which is fixed in the CVS version of automake.
Olly If you'd prefer not to use that you
"Jim" == Jim Meyering [EMAIL PROTECTED] writes:
Jim I've just looked into the failure of test pr19.
I don't understand.
pr19.test works fine for me.
Tom
"Hal" == Duston, Hal [EMAIL PROTECTED] writes:
Hal I am setting my package up with automake/autoconf, and want to
Hal provide gnu getopt if it is not available. I noticed
Hal AC_REPLACE_GNU_GETOPT in the info file, but it doesn't seem to be
Hal completely implemented. It is in automake, and
Hal That would be something I would install on my development system?
Hal Would I need to put it somewhere special so it doesn't go away if
Hal I upgrade automake? I would really rather do it the "proper" way
Hal if true. Or is get a copy the "proper" way? Still trying to get
Hal my brain
801 - 890 of 890 matches
Mail list logo