Re: [PATCH 03/72] t/test-lib.sh: add always-set NO_GETTEXT_POISON prerequisite

On Sun, Feb 20, 2011 at 22:55, Jonathan Nieder [off-list ref] wrote:

Ævar Arnfjörð Bjarmason wrote:

quoted

It's explicitly about tests that can't deal with poison, not
non-English. See this comment in patch  28/72:

    gettextize: git-commit "enter the commit message" message

    Gettextize the "# Please enter the commit message for your changes."
    message. Several tests in t7500-commit.sh and t7502-commit.sh assume
    that this message starts with a newline. Change the tests to to skip
    under GETTEXT_POISON=YesPlease.

Shouldn't the functional part (the starting newline, opening # marks)
be untranslatable then, to avoid making a trap for translators?  Such
a fix could happen at any time, and until then, we could skip those
tests in the non-English case.

I'm aiming to make this as minimal as possible, so I'm not going to
change the code to start constructing string buffers where we
previously had literals in this iteration.

And the trap is minimimal, sinc msgfmt --check will warn if the
structure of the newlines in the translation is different.

Since that codepath requires human intervention anyway (which is
generally slow), I can't imagine doing that would hurt runtime
performance enough to matter.

But I do get your point --- perhaps in some other circumstance we have
to rely on some intelligence by the translator for correct behavior.
So maybe it means something along the lines of

       test_set_prereq SANE_TRANSLATION

Well, the feature I have is to inject garbage into the gettext strings
in an effort to smoke out when I break the plumbing. So I think given
that functionality calling it NO_GETTEXT_POISON makes sense.

But I plan to add something to smoke other languages etc. later. Then
I might name that other prerequisite something else.