On Mon, Jan 11, 2021 at 03:47:34PM +0100, Ævar Arnfjörð Bjarmason wrote:
> As argued upthread and in 2/6 this test setup has outrun its
> usefulness.
>
> This series begins to remove it in a way that's very careful about not
> stepping on the toes of any other in-flight patches. In particular
> "master" isn't anywhere in the diff context here to avoid conflicts
> with the series to do that mass-replacement in t/.
>
> This series merges and tests cleanly in combination with the current
> "seen" branch.
>
> At the end we're left with a bunch of tests still using
> test_i18n{cmp,grep}, but I think that's preferable to having in-flight
> conflicts. It'll be trivial to search-replace those at some point in
> the future where we don't have another search-replacement series
> amending t/ in-flight.
>
> Ævar Arnfjörð Bjarmason (6):
> ci: remove GETTEXT_POISON jobs
> tests: remove support for GIT_TEST_GETTEXT_POISON
> tests: remove misc use of test_i18n{cmp,grep}
> tests: (almost) remove use of "test_i18ngrep !"
> tests: (almost) remove C_LOCALE_OUTPUT prerequisites
> tests: remove uses of GIT_TEST_GETTEXT_POISON=false
At the end of this patch series there are still a couple of remnants
left:
- t3406: has a comment about gettext poisoning.
- Makefile still has a warning about for the GETTEXT_POISON build
knob referring to GIT_TEST_GETTEXT_POISON and t/README.
I think both can be simply removed.
