This is the mail archive of the cygwin mailing list for the Cygwin project.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
Other format: | [Raw text] |
Brian Dessent wrote: > Dave Korn wrote: > >> I'm having trouble with the cygwin auto*tool package and the wrappers >> it puts around the different tool versions. I've looked in the >> following places > > I think that they key point is that for automake, the automatic version > guessing is no longer done, there are no wrapper scripts. I don't think > the WANT_ or FORCE_ stuff has any effect. Umm, it definitely *does* have an effect; see the differences it made to the examples I posted in the previous case. Is it possible I have some remnants of the old autoconf stuff lying around? I updated this morning and didn't see any problems, and nor does cygcheck report any, and a quick browse through /usr/share/autotools/ac-wrapper.pl shows it clearly tests and responds to WANT_ and FORCE_ env vars. > You should be able to just run e.g. "automake-1.4" if you want that > version. When you run "automake" you get whichever version is currently > set as the "alternative" -- see "man alternatives". You can run > "update-alternatives --config automake" to select which version the > generic "automake" should run. > > Note that for autoconf, there is still "version guessing" so you should > always be able to just run autoconf and get the right thing. It doesn't work OOTB. Here's a simple testcase: on my system, which is up-to-date as of this morning, and on which I've never ever run alternatives to change any settings, you can't build current binutils when you configure with --enable-maintainer-mode. ----------------------snip #!/bin/bash cd /tmp rm -rf autotooltest mkdir autotooltest cd autotooltest wget 'http://ftp.gnu.org/gnu/binutils/binutils-2.16.1.tar.bz2' tar xjf binutils-2.16.1.tar.bz2 mkdir obj cd obj /tmp/autotooltest/binutils-2.16.1/configure -v --enable-maintainer-mode 2>&1 | tee conf.log make all 2>&1 | tee build.log ----------------------snip ...which gets as far as configuring and building the bfd subdir, but then fails when it recurses into the binutils subdir, at the config.status --recheck stage, with this error: ----------------------snip creating ./config.status creating Makefile creating doc/Makefile creating po/Makefile.in creating config.h make[1]: Entering directory `/tmp/autotooltest/obj/binutils' cd /tmp/autotooltest/binutils-2.16.1/binutils && aclocal-1.4 cd /tmp/autotooltest/binutils-2.16.1/binutils && automake-1.4 --cygnus Makefile cd /tmp/autotooltest/binutils-2.16.1/binutils && autoconf FATAL ERROR: Autoconf version 2.50 or higher is required for this script make[1]: *** [/tmp/autotooltest/binutils-2.16.1/binutils/configure] Error 2 make[1]: Leaving directory `/tmp/autotooltest/obj/binutils' make: *** [all-binutils] Error 2 dk@espanola /> ----------------------snip I don't get that. I've got the default alternatives set, so IIUIC it should have selected the most recent autoconf, shouldn't it? I imagine this is intended to work with everything in the default installation state, but as you see it certainly doesn't WFM! > This seems to be mostly documented in Chuck's announcement here: > <http://www.cygwin.com/ml/cygwin-announce/2005-08/msg00059.html> Yeh, I did see that post. This is just the first time since last August that I've had to regenerate an autogenerated file. The use of alternatives, while alluded to, is not exactly documented... it would have helped if a sample command line was included, or at least the name of the package mentioned; anyone might think they could configure autoconf by running "update-alternatives --setup autoconf", for instance, which just exits silently without doing anything or reporting any kind of error. I'm sorry. I'm just confused. I've read the announcement, I've tried the simplest tests possible, as far as I can see automake and aclocal are controlled by the new alternatives system but autoconf is still way out there on its own somewhere, I really think the package is faulty. [ For the moment, I'm working around this by hand copying my changes from Makefile.am to Makefile.in, but it would be nice to resolve! ] cheers, DaveK -- Can't think of a witty .sigline today....
Attachment:
cygcheck.out.txt
Description: Text document
Attachment:
test-script.sh
Description: Binary data
-- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Problem reports: http://cygwin.com/problems.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |