RESOLVED FIXED 203418
[EWS] Status page should show compiler ERRORS
https://bugs.webkit.org/show_bug.cgi?id=203418
Summary [EWS] Status page should show compiler ERRORS
Brent Fulgham
Reported 2019-10-25 09:03:12 PDT
The EWS results presented when you click on a failure bubble currently gives you the option of looking at warnings, or looking at the entire log file (which can be tens of thousands of lines). We should also offer a filtered "errors" view that only shows the compile errors. These are generally the most important things to look at. Warnings are generally not as useful for figuring out why an EWS bot is unhappy.
Attachments
Current view (51.75 KB, image/png)
2019-10-25 09:04 PDT, Brent Fulgham
no flags
Patch (2.96 KB, patch)
2019-10-25 11:45 PDT, Aakash Jain
no flags
Patch (2.92 KB, patch)
2019-10-25 12:51 PDT, Aakash Jain
jbedard: review+
Screenshot - view after this change (259.16 KB, image/png)
2019-10-28 09:16 PDT, Aakash Jain
no flags
Brent Fulgham
Comment 1 2019-10-25 09:04:10 PDT
Created attachment 381929 [details] Current view Here's a screenshot showing the two options. A much better tool would be a summary of the compiler errors.
Brent Fulgham
Comment 2 2019-10-25 09:06:40 PDT
You might consider passing the output through Dino's old "filter-build-webkit" script, which did a good job of culling out boilerplate noise, and showing only the meaningful output from a compile.
Radar WebKit Bug Importer
Comment 3 2019-10-25 09:07:04 PDT
Aakash Jain
Comment 4 2019-10-25 11:45:22 PDT
Jonathan Bedard
Comment 6 2019-10-25 11:51:27 PDT
Comment on attachment 381950 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=381950&action=review > Tools/BuildSlaveSupport/ews-build/steps.py:747 > + is_whitespace = self.whitespace_re.search(line) is not None This temporary variable doesn't seem super useful > Tools/BuildSlaveSupport/ews-build/steps.py:754 > + map(self.errorReceived, self.error_context_buffer[:-1]) Why the -1? Couldn't we just map errorReceived to the whole buffer, which would include the line?
Brent Fulgham
Comment 7 2019-10-25 11:55:49 PDT
(In reply to Aakash Jain from comment #4) > Created attachment 381950 [details] > Patch Oh gosh! I'm so happy! :-)
Aakash Jain
Comment 8 2019-10-25 12:50:18 PDT
(In reply to Jonathan Bedard from comment #6) > Why the -1? Couldn't we just map errorReceived to the whole buffer, which would include the line? Changed in updated patch.
Aakash Jain
Comment 9 2019-10-25 12:51:06 PDT
Brent Fulgham
Comment 11 2019-10-25 13:44:38 PDT
Aakash Jain
Comment 12 2019-10-25 15:07:16 PDT
Aakash Jain
Comment 13 2019-10-26 08:48:40 PDT
Deployed on the production server.
Aakash Jain
Comment 14 2019-10-28 09:16:44 PDT
Aakash Jain
Comment 15 2019-10-28 10:57:15 PDT
*** Bug 201941 has been marked as a duplicate of this bug. ***
Note You need to log in before you can comment on or make changes to this bug.