On Thu, Feb 15, 2018 at 5:52 PM, Jeff King <p...@peff.net> wrote: > On Thu, Feb 15, 2018 at 05:39:28PM +0100, SZEDER Gábor wrote: > >> On Tue, Feb 13, 2018 at 11:22 PM, Jeff King <p...@peff.net> wrote: >> >> >> for i in "--perl --shell" "-s --python" "--python --tcl" "--tcl --perl"; >> >> do >> >> test_expect_success "more than one quoting style: $i" " >> >> - git for-each-ref $i 2>&1 | (read line && >> >> - case \$line in >> >> - \"error: more than one quoting style\"*) : happy;; >> >> - *) false >> >> - esac) >> >> + test_must_fail git for-each-ref $i 2>err && >> >> + grep '^error: more than one quoting style' err >> > >> > I suspect in the long run this ought to be test_i18ngrep, but since it's >> > not localized yet, it makes sense to stop here with this patch. >> >> I thought 'git for-each-ref' is plumbing and that means that it >> shouldn't be localized. > > I always assumed stderr was mostly fair game, even for plumbing. At any > rate, I'm willing to ignore the issue until somebody actually proposes a > patch to translate it.
Ah, OK, will keep that in mind. Anyway, the first GETTEXT_POISON build will fail if that error string ever gets translated but the test doesn't get updated.
