Bug 91465 – nrwt: move per-test result output into printing.py

• Status: RESOLVED
• Resolution: FIXED
• Component: New Bugs

[Dirk Pranke]

nrwt: move per-test result output into printing.py

[Dirk Pranke]

Created attachment 152685 [details]
Patch

[Dirk Pranke]

Created attachment 152686 [details]
add comment about removing dead code

[Ojan Vafai]

Comment on attachment 152686 [details]
add comment about removing dead code

It looks like if you choose trace-everything (which we do on the bots, right?) we now print something for skipped tests where we didn't before. Was that intentional?

[Dirk Pranke]

(In reply to comment #3)
> (From update of attachment 152686 [details])
> It looks like if you choose trace-everything (which we do on the bots, right?) we now print something for skipped tests where we didn't before. Was that intentional?

We don't run with --print trace-everything on the bots. 

However, even if we did, nothing would get printed for skipped tests. update_summary_with_result is only called from handle_finished_test, and so it's only called for tests that are actually run. 

Which means that the 'if result.type == test_expectations.SKIP:' branch is never True :). So I'll delete it. Good catch.

[Dirk Pranke]

Created attachment 152797 [details]
remove unused branch in update_summary_with_result()

[Ojan Vafai]

Comment on attachment 152797 [details]
remove unused branch in update_summary_with_result()

View in context: https://bugs.webkit.org/attachment.cgi?id=152797&action=review

> Tools/Scripts/webkitpy/layout_tests/controllers/manager.py:-981
> -        if result.type == test_expectations.SKIP:

Maybe mention in the ChangeLog why it's safe to remove this?

[Dirk Pranke]

(In reply to comment #4)
> (In reply to comment #3)
> > (From update of attachment 152686 [details] [details])
> > It looks like if you choose trace-everything (which we do on the bots, right?) we now print something for skipped tests where we didn't before. Was that intentional?
> 
> We don't run with --print trace-everything on the bots. 
> 
> However, even if we did, nothing would get printed for skipped tests. update_summary_with_result is only called from handle_finished_test, and so it's only called for tests that are actually run. 
> 
> Which means that the 'if result.type == test_expectations.SKIP:' branch is never True :). So I'll delete it. Good catch.

Ack, I'm wrong. if we skip a test because it was a reftest and --no-ref-tests was on the command line, then we will pass SKIP as a result. I will revert that part of the code, although this seems like a separate bug (we should just be skipping the reftests up front).

[Dirk Pranke]

note that I think it's okay to print this case in --print trace-everything.

[Dirk Pranke]

Committed r122897: <http://trac.webkit.org/changeset/122897>