On Tue, Jan 12, 2021 at 1:11 AM Akim Demaille <a...@lrde.epita.fr> wrote: > > ... > > Would you agree to try to same test with the current git version > of Bison? It's available here: > > https://www.lrde.epita.fr/~akim/private/bison/bison-3.7.4.254-a0905.tar.gz > https://www.lrde.epita.fr/~akim/private/bison/bison-3.7.4.254-a0905.tar.lz > https://www.lrde.epita.fr/~akim/private/bison/bison-3.7.4.254-a0905.tar.xz > > There is no known issue in this tarball, it's just as good as a release, > except that it includes unfinished new features. > > Build it, and check it. Test 263 in 3.7.4 is test 271 in this tarball. > Once 271 failed, please run: > > ./src/bison -Tcex -Wcex ./tests/testsuite.dir/271/input.y >/tmp/271.log 2>&1 > > Compress and please send /tmp/271.log. It is probably very long, > instead of sending it, maybe you can leave it somewhere on the > Internet.
Yes, sure. Give me a few minutes. If it matters, Alpine 3.10 is a Musl and BusyBox machine. Sometimes it produces unexpected results because Rich is a little more strict about following standards. For example, a memcpy(NULL, ....) will succeed with Glibc but crash Musl because the standard does not allow a NULL pointer there. Jeff