Re: Please test autoreconf from zack/autoreconf-exclude branch
"Zack Weinberg" <[email protected]> Mon, 09 Mar 2026 15:42:36 -0400
| Newsgroups | gmane.comp.sysutils.autoconf.general |
|---|---|
| Message-ID | <[email protected]> |
On Mon, Mar 9, 2026, at 2:08 PM, Zack Weinberg wrote: > On Fri, Mar 6, 2026, at 11:05 AM, Ross Burton wrote: >> diffstat-1.68 fails to autoreconfig with this release. > > Applying autoreconf from *any* released version of GNU autoconf to this tarball > will produce a configure script that doesn't work, or works only by > luck. As of commit 4d966682 (just now) GNU autoconf will error out on this configure.in: /tmp/diffstat-1.68 $ ~/projects/gnu/autoconf/_build/tests/autoconf configure.in:86: error: AC_OUTPUT takes at most three arguments. This configure script cannot be compiled by this version of Autoconf. configure.in:86: the top level autom4te: error: /usr/bin/m4 failed with exit status: 1 The defensive measure only covers one of the several places where Thomas Dickey's forked autoconf diverges from official GNU autoconf but I hope it will be sufficient to prevent people from wasting a lot of time trying to figure out why they can't persuade GNU autoconf to produce a working configure script from this kind of configure.in. Thomas: I personally would be willing to work with you to see your changes and fixes merged back in to official GNU autoconf, to the extent this is possible (for instance, your @[email protected] mode would have to be expressed using a new AC_CONFIG_something macro, as it is far too late to un-deprecate the arguments to AC_OUTPUT). From what you've written about it, I get the impression you would rather not be bothered, so I won't bring it up again unless you tell me you do want to make that happen. zw