Handling of extra macros

2005-12-01 Thread Ross Boylan
I've seen conflicting advice about how to handle extra macros, whether
developed locally or from the archive.  What's the best way to get them
into configure?  I think this reduces to the question of how to make
aclocal aware of them.

1. AC_CONFIG_MACRO_DIR doesn't work, according to prior posts on this
list, my tests, and, ambiguously, the manual (which refers to future
versions).

2. The man page for aclocal on my (Debian) system says  At startup,
aclocal scans all the .m4 files it can  find,  looking  for macro
definitions.  Then it scans configure.in.  Any mention of one of the
macros found in the first step causes that macro, and any macros it in
turn requires, to be put into aclocal.m4.  This doesn't seem to be the
case, even if the file is in the same directory as configure.ac.  Or one
could say the quoted description is true, but aclocal can't find macros
in the project directory.  Either way, it doesn't work.

3. Other advice says to stick the macros in aclocal.m4.  I did, but they
got overwritten the first time aclocal ran.

4. If I put the macros in a directory and do
   aclocal -I thedirectory
things work, but this seems to be asking for trouble.

5. The autoconf manual has no index entry for aclocal.  There is one for
aclocal.m4, and the text refers to aclocal.  Is aclocal deprecated?

I have Debian autoconf 2.59a-7 and am looking at the manual for 2.59,
dated 5 Nov 2003 (direct from GNU).  I've also been looking at the
somewhat dated Autoconf, Automake, and Libtool book.  I'm not using
automake.
-- 
Ross Boylan  wk:  (415) 514-8146
185 Berry St #5700   [EMAIL PROTECTED]
Dept of Epidemiology and Biostatistics   fax: (415) 514-8150
University of California, San Francisco
San Francisco, CA 94107-1739 hm:  (415) 550-1062



___
Autoconf mailing list
Autoconf@gnu.org
http://lists.gnu.org/mailman/listinfo/autoconf


Re: Handling of extra macros

2005-12-01 Thread Ross Boylan
On Thu, 2005-12-01 at 21:38 +0100, Alexandre Duret-Lutz wrote:
  RB == Ross Boylan [EMAIL PROTECTED] writes:
 
  RB 5. The autoconf manual has no index entry for aclocal.  There is one for
  RB aclocal.m4, and the text refers to aclocal.  Is aclocal deprecated?
 
 aclocal is part of Automake, not autoconf.  The Automake manual
 documents aclocal and explains how to handle local macros properly.
Thank you.  For the record, the file acinclude.m4 is automatically
included.

It seems odd that a routine part of using autoconf requires automake.  I
realize technically aclocal is not required to use autoconf, but my
understanding is that it's use is routine.




___
Autoconf mailing list
Autoconf@gnu.org
http://lists.gnu.org/mailman/listinfo/autoconf


Problems running autoscan

2004-12-02 Thread Ross Boylan
$ autoscan 
autom4te: configure.ac: no such file or directory
autoscan: /usr/bin/autom4te failed with exit status: 1

This is a puzzling message, since the purpose of autoscan, as I
understand it, is to produce a file which, if blessed by me, becomes
configure.ac.  So why would it complain that configure.ac is missing?

Does this just indicate that autoscan does additional stuff for people
who have configure.ac, or is something wrong?

$ autoscan --version
autoscan (GNU Autoconf) 2.59
Written by David J. MacKenzie and Akim Demaille.

Running on Debian GNU/Linux.  It may be relevant that the project has
parts in subdirectories.


___
Autoconf mailing list
[EMAIL PROTECTED]
http://lists.gnu.org/mailman/listinfo/autoconf


kdelibs build problem: possibly undefined macro: AC_CHECK_MKDTEMP

2003-06-06 Thread Ross Boylan
I am trying to build KDE on a Debian GNU/Linux testing system (build
from cvs head).  I keep running into problems that I think are related
the versions of the auto* tools I'm using.

Do I need particular versions of autoconf to work with particular
versions of automake, for example?  How do I find which ones are
consistent?

And how can I figure out what is going on with errors like the
following, while trying to build kdelibs:
make -k -f Makefile.cvs
This Makefile is only for the CVS repository

*** Creating acinclude.m4
*** Creating list of subdirectories
*** Creating configure.in
*** Creating aclocal.m4
*** Creating configure
configure.in:105: error: possibly undefined macro: AC_CHECK_MKDTEMP
  If this token and others are legitimate, please use m4_pattern_allow.
  See the Autoconf documentation.

Some version info
autoconf (GNU Autoconf) 2.57
automake (GNU automake) 1.7.5
gcc.real (GCC) 3.3 (Debian)
autotools-dev  20030519.1 (provides config.guess and config.sub)

I had some lower version number tools installed earlier, and managed
to get past some problems by removing them.  I believe I only have one
each of autconf and automake now, though several flavors of gcc are on
the system.  Some of the KDE scripts explicitly search for particular
versions of the tools.

Thanks.





Re: kdelibs build problem: possibly undefined macro: AC_CHECK_MKDTEMP

2003-06-05 Thread Ross Boylan
On Thu, Jun 05, 2003 at 01:33:29PM -0700, Ross Boylan wrote:
 I am trying to build KDE on a Debian GNU/Linux testing system (build
 from cvs head).  I keep running into problems that I think are related
 the versions of the auto* tools I'm using.
 
 Do I need particular versions of autoconf to work with particular
 versions of automake, for example?  How do I find which ones are
 consistent?
 
 And how can I figure out what is going on with errors like the
 following, while trying to build kdelibs:
 make -k -f Makefile.cvs
 This Makefile is only for the CVS repository
 
 *** Creating acinclude.m4
 *** Creating list of subdirectories
 *** Creating configure.in
 *** Creating aclocal.m4
 *** Creating configure
 configure.in:105: error: possibly undefined macro: AC_CHECK_MKDTEMP
   If this token and others are legitimate, please use m4_pattern_allow.
   See the Autoconf documentation.

OK, this problem was an odd side effect of mixing cvs and cvsup, which
froze some files at older versions.  The macro is defined in some
other KDE files.

I am still interested in answers on the general questions raised at
the top of the message.