We should extract the logic to build and run the lldbwebkit tests from test-webkitpy to a new script, say test-lldbwebkit, to avoid mixing the lldbwebkit tests with the webkitpy tests.
One of the reasons is that we’d need to run these tests for WebCore changes, while webkitpy tests only need to run for webkitpy changes.
Running tests as part of test-webkitpy also creates difficulties in settings where we dont’t have a WebKit build in the first place.
Created attachment 348663 [details]
I will teach the EWS and build.webkit.org build bots how to run test-lldb-webkit in a subsequent patch/bug.
Filed bug #189205 and bug #189206 to teach the build bot and EWS how to run the LLDB WebKit tests.
(In reply to Daniel Bates from comment #2)
> Created attachment 348663 [details]
> I will teach the EWS and build.webkit.org build bots how to run
> test-lldb-webkit in a subsequent patch/bug.
I don’t have time at the moment to do this. Maybe someone else can?
Created attachment 355684 [details]
(In reply to Jonathan Bedard from comment #6)
> Created attachment 355684 [details]
How is this different than my patch?
(In reply to Daniel Bates from comment #7)
> (In reply to Jonathan Bedard from comment #6)
> > Created attachment 355684 [details]
> > Patch
> How is this different than my patch?
I missed that, sorry. I saw comment 4 but not the patch.
Comment on attachment 348663 [details]
View in context: https://bugs.webkit.org/attachment.cgi?id=348663&action=review
> + if os.path.isdir(lldb_python_directory):
Shouldn't we use the host object here to conform to webkitpy's idioms?
Can we add a --root option here? (matches with run-webkit-tests and run-api-tests)
> + self.printer.write_update("Checking autoinstalled packages ...")
We shouldn't need to do this for lldb tests
> +class _Loader(unittest.TestLoader):
I feel like this should be class Loader(...) if we're allowing this to be imported, it shouldn't start with '_'