On 11/17/10 07:35, Eric Blake wrote: > On 11/17/2010 08:11 AM, Bruce Korb wrote: >>> OK. I don't see where it should come from in Autoconf nor Automake. >>> Any case the package at hand contains m4_esyscmd in configure.ac that >> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ >>> contains a buggy sed script? If not, please state exact versions of >>> autotools programs used, and show autoreconf -v output. >> >> There's a really good chance of that. >> >> AC_INIT([GNU libposix], >> m4_esyscmd([./git-version-gen .tarball-version]), >> [bug-gnu...@gnu.org]) >> >> $ for f in autoconf automake libtool autoreconf ; do $f --version | head -1 >> ; done >> autoconf (GNU Autoconf) 2.65 >> automake (GNU automake) 1.11.1 >> ltmain.sh (GNU libtool) 2.2.6 >> autoreconf (GNU Autoconf) 2.65 > > "git grep 'sed.* -q'" in automake, autoconf, gnulib, and libtool doesn't > turn up any culprits. > > Is it worth installing a temporary sed wrapper earlier in your PATH that > outputs debugging information such as the process id of its parent if it > detects that one of the arguments starts with -q?
Next time I run into it. I unwound everything I was doing and started reapplying my changes. Before I get to where it breaks again, I am having to figure out how to "git push" without tripping over "fast-forward", whatever that is.
