On 10 February 2018 at 09:52, Axel Beckert <a...@debian.org> wrote: > Control: tag -1 + confirmed > > Hi Matthias, > > Matthias Klose wrote: > > make check-TESTS check-local > > make[5]: Entering directory '/<<PKGBUILDDIR>>' > > echo ./tests/*.el ./tests/interactive/*.el | abs_srcdir=/<<PKGBUILDDIR>> > > srcdir=. TERM=unknown VALGRIND="" EMACSPROG="" xargs /usr/bin/perl > > ./tests/run-lisp-tests.pl > > make[6]: Entering directory '/<<PKGBUILDDIR>>' > > Error opening terminal: unknown. > > Zile failed to run test `backward-char' with error code 256 > > Error opening terminal: unknown.[...] > > This probably happens for a long time, but my try to make the build > not garble my terminal during package build testing with "cat -v" had > covered dh_auto_test's exit code. The latter is fixed now and hence > test suite failures come to surface as they should > > Anyway, confirmed locally with > > $ env TERM=foobar debuild -uc -us > > > seen on the Ubuntu buildds. > > Interestingly not on Debian's buildds (except hurd-i386) nor in > pbuilder (which AFAIK doesn't pass $TERM). > > Will add a fix. >
I just want to check, do you nonetheless consider this an upstream bug? Should I recognise the value "unknown" as being equivalent to unset TERM? The problem for me is that I don't want to override any setting from the user: for example, it is nice if a user who builds from source gets the tests run against the terminal they happen to be using. -- https://rrt.sc3d.org