Re: AC_TYPE_INT8/16/32/64_T fail

2010-02-06 Thread Eric Blake
According to Daily, Jeff A on 2/6/2010 3:22 PM: > The AC_TYPE_INT*_T macros worked for me with 2.64, but now they don't with > 2.65. This is a known regression, fixed here: http://git.sv.gnu.org/cgit/autoconf.git/commit/?id=83ee5bc4 I hope to release autoconf 2.66 before the end of this month, w

AC_TYPE_INT8/16/32/64_T fail

2010-02-06 Thread Daily, Jeff A
The AC_TYPE_INT*_T macros worked for me with 2.64, but now they don't with 2.65. The AC_TYPE_UINT*_T macros still work. I have verified that reverting back to 2.64 does work for me. I don't have the output of make check to provide, but I do have the output from my config.log for my software w

Re: Weird interactions between AC_PROG_CC and AC_PROG_F77

2010-02-06 Thread Stefano Lattarini
At Saturday 06 February 2010, Ralf Wildenhues wrote: > Hello Stefano, Hello Ralf. Thanks for your quick answer. > * Stefano Lattarini wrote on Sat, Feb 06, 2010 at 04:14:45PM CET: > > I encountered the following strage behaviour in the latest stable > > autoconf (2.65, Debian package version 2.6

Re: Weird interactions between AC_PROG_CC and AC_PROG_F77

2010-02-06 Thread Ralf Wildenhues
Hello Stefano, * Stefano Lattarini wrote on Sat, Feb 06, 2010 at 04:14:45PM CET: > I encountered the following strage behaviour in the latest stable > autoconf (2.65, Debian package version 2.65-2), and it really seems > a bug to me: Of the AC_PROG_{CC,CXX,F77,FC} macros, only the first one in t

Weird interactions between AC_PROG_CC and AC_PROG_F77

2010-02-06 Thread Stefano Lattarini
Hello everybody. I encountered the following strage behaviour in the latest stable autoconf (2.65, Debian package version 2.65-2), and it really seems a bug to me: $ cat > configure.ac < configure.ac <

Re: Creating autotest testsuite gets stuck if package.m4.in contains unresolved substitutions

2010-02-06 Thread Eric Blake
According to Niels Lohmann on 2/5/2010 3:50 AM: > This all works, but if (by accident), a check in configure.ac is deleted, say > than one for WGET, then tests/package.m4 ends up with an unresolved > substitution: > > m4_define([WGET], [...@wget@]) Yes, any attempt to expand WGET in m4 at

Re: Creating autotest testsuite gets stuck if package.m4.in contains unresolved substitutions

2010-02-06 Thread Ralf Wildenhues
Hello Niels, * Niels Lohmann wrote on Fri, Feb 05, 2010 at 11:50:10AM CET: > we have a project where we use the configure script to search for some > helper tools and write the paths to the package.m4 for the autotest > test suite. We do this with > > configure.ac: > AC_PATH_PROGS(WGET, [wg

Re: autoconf 2.65: AC_PROG_F77 code executed though in a test that is always false

2010-02-06 Thread Ralf Wildenhues
Hello Vincent, thanks for the report. * Vincent Lefevre wrote on Fri, Feb 05, 2010 at 02:51:25PM CET: > I can reproduce it under Mac OS X (with autoconf 2.65). > > Note that this is a regression. On a Debian/stable (lenny) machine > with autoconf 2.61-8, there is no such problem. [...] > On 2010