[RFC][Bug #111914]selftest options to fail if there are missing dependencies, known-failures, etc
Martin Pool
mbp at sourcefrog.net
Mon Jul 30 19:43:55 BST 2007
On 7/29/07, Daniel Watkins <D.M.Watkins at warwick.ac.uk> wrote:
> Daniel Watkins wrote:
> > Attached is a preliminary fix for bug #111914.
> The above is actually true of this post.
That looks ok. You should put the string '[merge]' in the mail
subject so that Bundle Buggy catches it and your patch isn't lost.
Does everything pass in bzr.dev with this merged?
You should probably add a note in the testing section of HACKING
explaining this.
It looks like this needs the finer-grained test results I posted about
a while ago to be most useful: we want --strict to fail if there are
known failures or missing dependencies, but not because of OS
limitations. But if it passes at present I don't mind putting it in
as preparation.
> @@ -2511,7 +2513,8 @@
> matching_tests_first=first,
> list_only=list_only,
> random_seed=randomize,
> - exclude_pattern=exclude
> + exclude_pattern=exclude,
> + strict=strict
> )
> finally:
> if benchfile is not None:
>
To make later updates easier it's good to include a trailing comma here.
> @@ -2206,7 +2207,12 @@
> suite = filter_suite_by_re(suite, pattern, exclude_pattern,
> random_order)
> result = runner.run(suite)
> - return result.wasSuccessful()
> + success = result.wasSuccessful()
> +
> + if strict and (result.known_failure_count or result.unsupported):
> + success = False
> +
> + return success
I wonder if evaluation of this should be moved into the result through
e.g. result.wasStrictlySuccessful()? Then as we add more results we
shouldn't need to change this method.
> def selftest(verbose=False, pattern=".*", stop_on_failure=True,
> @@ -2217,7 +2223,8 @@
> matching_tests_first=None,
> list_only=False,
> random_seed=None,
> - exclude_pattern=None):
> + exclude_pattern=None,
> + strict=False,):
> """Run the whole test suite under the enhanced runner"""
> # XXX: Very ugly way to do this...
> # Disable warning about old formats because we don't want it to disturb
And if you have a comma the paren should be on the following line.
--
Martin
More information about the bazaar
mailing list