tests should use C locale

[Jelmer Vernooij]

the tests currently rely on particular strings not being translated. for example, I have LC_ALL=de_DE and I now see some failures:

Traceback (most recent call last):
  File "/home/jelmer/src/bzr/bzrlib/tests/blackbox/test_exceptions.py", line 69, in test_utf8_default_fs_enc
    self.assertContainsRe(out, "^Created a standalone tree .*$")
AssertionError: pattern "^Created a standalone tree .*$" not found in
"""\
Ein standalone tree wurde erstellt (Format: 2a)
"""

We should probably force the locale to be C during the tests.

[Alexander Belchenko (bialix)]

Jelmer Vernooij пишет:
> We should probably force the locale to be C during the tests.

In QBzr and bzr-explorer we explicitly disable i18n during tests. You
don't have to force C locale, just disable i18n.

[Jelmer Vernooij (jelmer)]

> Jelmer Vernooij пишет:
> > We should probably force the locale to be C during the tests.
> In QBzr and bzr-explorer we explicitly disable i18n during tests. You
> don't have to force C locale, just disable i18n.
Uhm, isn't that the same thing, at least where translations are concerned?

[Alexander Belchenko (bialix)]

Jelmer Vernooij пишет:
>> Jelmer Vernooij пишет:
>>> We should probably force the locale to be C during the tests.
>> In QBzr and bzr-explorer we explicitly disable i18n during tests. You
>> don't have to force C locale, just disable i18n.
> Uhm, isn't that the same thing, at least where translations are concerned?

I think it's different things. Forcing C locale will affect how
unicode tests behave in bzrlib, but disabling i18n just forces English
messages, but does not affect unicode.

--
All the dude wanted was his rug back

[Jelmer Vernooij (jelmer)]

Am 31/03/12 15:00, schrieb Alexander Belchenko:
> Jelmer Vernooij пишет:
>>> Jelmer Vernooij пишет:
>>>> We should probably force the locale to be C during the tests.
>>> In QBzr and bzr-explorer we explicitly disable i18n during tests. You
>>> don't have to force C locale, just disable i18n.
>> Uhm, isn't that the same thing, at least where translations are concerned?
> I think it's different things. Forcing C locale will affect how
> unicode tests behave in bzrlib, but disabling i18n just forces English
> messages, but does not affect unicode.
I agree, we shouldn't override the terminal encoding.

However, the locale does matter for more than just translations. For
example, it determines how dates are formatted.

Cheers,

Jelmer