Re: [Monotone-devel] Cygwin tests completed

[Justin Patrin]

On 2/7/07, Matthew Gregan <address@hidden> wrote:
> At 2007-02-07T10:45:56-0800, Justin Patrin wrote:
> > I've completed running the monotone test suite on cygwin (it took a
> > *long* time) and there are 7 failures. Most of these should likely be
> > set to expected fail on cygwin. Here's the list of failures:
>
> > 26 importing_files_with_non-english_names        FAIL (line 24)
>
> This one should be fixable...

Still failing currently.

> > 148 empty_environment                             FAIL (line 25)
>
> This test needs tweaking on some platforms--take a look at
> tests/empty_environment/__driver__.lua for some existing examples.
> Actually, I see it already has a tweak for cygwin, so maybe it just needs
> fixing or updating.
>
> The problem is that there are platforms that are unable to find the expected
> runtime files if we clear the environment out completely, so we have a nasty
> workaround where we copy the required runtime bits into the test directory
> before trying to run with an empty environment.

Matthew helped me fix this one.

> > 223 fail_cleanly_on_unreadable__MTN_options       FAIL (line 12)
> > 277 fail_cleanly_on_unreadable_db                 FAIL (line 36)
> > 288 database_is_closed_on_signal_exit             FAIL (line 24)

These are also fixed.

> > 362 log_quits_on_SIGPIPE                          FAIL (line 78)
>
> These are all skipped on Windows already, but Cygwin is probably slipping
> past the OS test.  We might be able to get them to work properly if Cygwin's
> POSIX emulation is sufficient, but I'd need to look at the test logs.

This one should be working but is not and needs some more work to fix it.

> > 393 test_a_merge_8                                FAIL (line 29)
>
> This one should be okay now, it's a new test that is expected to fail but
> hadn't been marked as xfailed initially.

--
Justin Patrin