It used to be the case that API tests would stream their output before timing out. After r230998, this output was cached and only printed after a test timed out.
Created attachment 339025 [details] Patch
Comment on attachment 339025 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=339025&action=review > Tools/Scripts/webkitpy/api_tests/runner.py:108 > + self.printer.writeln(source) Does this output with timestamps? I really don't want to see that in the streamed results. Because it would make copying & pasting multiple lines of output hard. e.g. for dumping render tree, etc... > Tools/Scripts/webkitpy/api_tests/runner.py:159 > + status = Runner.STATUS_FAILED Sounds like this is really STATUS_RUNNING? > Tools/Scripts/webkitpy/api_tests/runner.py:166 > + if status != Runner.STATUS_DISABLED: > + server_process.start() Why is this code change needed? > Tools/Scripts/webkitpy/api_tests/runner.py:177 > + self._caller.post('log', None, None, stderr_line[:-1]) Name arguments to which you're specifying None. Otherwise, the reader of this code can't tell what's being set to None. > Tools/Scripts/webkitpy/api_tests/runner.py:183 > + self._caller.post('log', None, None, stdout_line[:-1]) Ditto.
Comment on attachment 339025 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=339025&action=review > Tools/Scripts/webkitpy/api_tests/runner.py:52 > + self._has_logged_for_test = True Why this instance variable initially set to True?
Comment on attachment 339025 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=339025&action=review >> Tools/Scripts/webkitpy/api_tests/runner.py:52 >> + self._has_logged_for_test = True > > Why this instance variable initially set to True? This is basically a cosmetic difference. False would mean an empty line between the output 'Running tests', true means there is no empty line. I thought that not having the empty line looked better. >> Tools/Scripts/webkitpy/api_tests/runner.py:108 >> + self.printer.writeln(source) > > Does this output with timestamps? I really don't want to see that in the streamed results. > Because it would make copying & pasting multiple lines of output hard. > e.g. for dumping render tree, etc... This would output timestamps in verbose mode, yes. It sounds like that isn't desirable, I'll change it. >> Tools/Scripts/webkitpy/api_tests/runner.py:166 >> + server_process.start() > > Why is this code change needed? Because we really shouldn't have been starting the process if the test was disabled in the first place. This code change isn't needed, exactly (the old way did work), but it is better because it doesn't spawn an extra process needlessly.
Created attachment 339123 [details] Patch
Comment on attachment 339123 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=339123&action=review > Tools/Scripts/webkitpy/api_tests/runner.py:100 > + source = '' if self._num_workers == 1 else source + ' ' I wanted to call out what changed here, base on Ryosuke's request in comment 2: 'Does this output with timestamps? I really don't want to see that in the streamed results.' The new patch will not print timestamps unless specifically asked to do so. It will also not print the worker information if only one worker is used. This should make logs easy to copy paste.
Comment on attachment 339123 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=339123&action=review > Tools/Scripts/webkitpy/api_tests/runner.py:53 > + self._has_logged_for_test = True We should add a comment here saying this would suppress an empty line between "Running tests" and the first test's output. > Tools/Scripts/webkitpy/layout_tests/views/metered_stream.py:57 > + self._print_timestamps = verbose if print_timestamps is None else print_timestamps It's a bit odd that this falls back to verbose but I guess changing it everywhere would be a lot of code changes.
Created attachment 340142 [details] Patch for landing
Comment on attachment 340142 [details] Patch for landing Clearing flags on attachment: 340142 Committed r231679: <https://trac.webkit.org/changeset/231679>
All reviewed patches have been landed. Closing bug.
<rdar://problem/40148803>