Re: utf8_tests.py failing (just me?)

John Szakmeister <john@szakmeister.net> writes:

> If I 'export LC_ALL=en_US.ISO8859-1', then the UTF-8 tests pass. I'm not
> sure what's happening, but it appears as though the locale really isn't
> changed when the command line client is launched (it appears that
> utf8_tests.py attempts to set the locale to "en_US.ISO8859-1" before
> executing the tests). I created a small test script and found that the
> environment doesn't appear to be propagated.

I think that's standard on Unix, calling setlocale() changes the
locale of the current process but doesn't change the environment, so
the setlocale() doesn't affect any subsequent subprocesses.

My guess is that the people seeing failures are those who a) are using
a non-8859-1 locale, and b) have an 8859-1 en_US locale available.
Previously the failing tests were skipped as the locale names didn't
match (different case), now the test is run and the non-propogation
causes the failure. The test script should probably do a setenv() on
LC_CTYPE as well as the setlocale().

-- 
Philip Martin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org