Re: How to pinpoint aclocal failure

[Yang Tse]

2008/7/30, Ralf Wildenhues wrote:

> aclocal endless loops are a pain to debug.  If you can go back to a
> known-good version (of configure.ac and *.m4 files), you may be able to
> bisect.

One of the many things that have already been tried is to comment out
the new macro invocations leaving in place the new macro definitions.
The result has been that aclocal 'finds' all the defined macros as
axpected, and also as expected those with commented out invocations
are not 'seen', but when afterwards aclocal calls autom4te and this
calls m4 it bombs out with the same failure as if the new macro
invocations had not been commented out.

BTW aclocal is 'finding' 232 macros scanned from all packages, and
'seeing' 165 macros which are actually used. Are this numbers common
or are this already too high ?

>  Otherwise I remember that there have been some issues in the
> interaction of Libtool or Autoconf macros.  You may be able to find
> clues when searching the mailing list archives for it.

I've already did that, but I'll search even more.

> Otherwise, can you make a tarball or VCS-repository available with which
> the issue may be reproduced?

It can be reproduced on Solaris systems with gm4 1.4.5 autoconf 2.61
automake 1.10 and libtool 1.5.24 installed issuing the following 3
commands:

cvs -d :pserver:address@hidden:/cvsroot/curl co curl
cd curl
./buildconf

On other systems it does not fail. But on Solaris it repeatably fails with:

/opt/csw/bin/gm4: memory exhausted
autom4te: /opt/csw/bin/gm4 failed with exit status: 1
aclocal: autom4te failed with exit status: 1

-- 
-=[Yang]=-