... imposed a dependency on autoconf 2.71. Such is the way of the world, but I suggest that's worth a mention in README.git, like the dependency on automake 1.16.1 already gets, because there's no released version of eg Debian with an autoconf
that new. The (unreleased) Debian Bookworm builds of autoconf and automake are (the first ones that are) new enough and still install on Debian Stretch. That's nice but...
> for people to try out who don't want to go through the process of bootstrapping
from a Git
... while I don't like to depend, I do struggle with this part ~every release. Today for instance I couldn't make clean or make distclean to get my existing git work area to build with those new autotools. Both those commands and ./bootstrap
failed here:
autoreconf: running: aclocal -I m4 --force -I m4
configure.ac:43: error: AM_COND_IF: no such condition "GL_COND_OBJ_GETLOADAVG"
/usr/share/aclocal-1.16/cond-if.m4:23: AM_COND_IF is expanded from...
m4/gnulib-comp.m4:63: gl_INIT is expanded from...
configure.ac:43: the top level
autom4te: error: /usr/bin/m4 failed with exit status: 1
aclocal: error: /usr/bin/autom4te failed with exit status: 1
autoreconf: error: aclocal failed with exit status: 1
./bootstrap: autoreconf failed
martind@sirius:~/download/make.aside$
After cloning fresh worked, I realized that I should have tried moving the gnulib subdirectory aside as well as the m4 one. That worked, so I'm doubly ready to try the recent bout of new code. It wasn't any inconvenience to throw away my work
area and start again, but I don't like to think how much time I wasted before resorting to that. I wouldn't call that egregious and I was, after all, able to solve it myself, but perhaps having the error messages now appear in a web search will help someone
else. Guess I just haven't found the bit of the documentation that tells me what I should have typed instead of make distclean.
From: Bug-make <bug-make-bounces+martin.dorey=hds.com@gnu.org> on behalf of Paul Smith <psmith@gnu.org>
Sent: Saturday, September 3, 2022 15:56
To: bug-make@gnu.org <bug-make@gnu.org>
Subject: New release of GNU make
***** EXTERNAL EMAIL *****
Hi all.
I would like to work on putting out a new release of GNU make. I've
fixed, or merged patches for (thanks especially to Dmitry) most of the
issues that I had in mind for this release, except one: I'd really like
to fix the problem we currently have with handling fatal signals (like
SIGINT), because die() is invoked in a signal handler context and it
does all sorts of things that are not valid in a signal handler.
I have a partial solution to this but it runs up against the age-old
problem of trying to both wait for data on a file descriptor and also
wait for a child to exit, reliably, in a POSIX system. I'm thinking
about ways to manage this.
But in the meantime I wanted to let people know in case they wanted to
point out particularly egregious bugs that should absolutely be looked
at before the release (obviously any regressions are most important).
As always I'll be creating release candidates and announcing them here
for people to try out who don't want to go through the process of
bootstrapping from a Git workspace.