Bug 153941

Summary: [GTK] visible-control-characters tests should not be passing
Product: WebKit Reporter: Michael Catanzaro <mcatanzaro>
Component: WebKitGTKAssignee: Nobody <webkit-unassigned>
Status: NEW ---    
Severity: Normal CC: bugs-noreply, mcatanzaro
Priority: P2    
Version: Other   
Hardware: PC   
OS: Linux   

Description Michael Catanzaro 2016-02-06 08:55:09 PST
The following tests are sometimes passing for the last couple of months:

  fast/text/control-characters/visible-control-characters-1.html
  fast/text/control-characters/visible-control-characters-2.html
  fast/text/control-characters/visible-control-characters-2.xhtml
  fast/text/control-characters/visible-control-characters-3.html
  fast/text/control-characters/visible-control-characters-3.xhtml

It's bad because these are expected mismatch tests, they should never pass.

The first pass (known bad commit) is r192730, but that commit is probably not to blame. Then there are many flaky runs. Then r192779 is the last known bad commit, and then test consistently fails (good) for a while, so it was fixed for a range of commits. The next known pass (bad) is r193641, and it's flaky from then on.

So we know:

* Tests broken shortly before r192730, probably exclusive
* Tests fixed shortly after r192779, exclusive
* Tests broken again sometimes before r193642, exclusive

Unfortunately there was some problem with the bot, and the last results we have before r193641 are for r193617, which is a rather big range of commits.

Since the tests were broken, fixed, then broken again, I at first suspected r192727, shortly before r192730. However that was rolled out quickly in r192731, but the tests continued to pass (bad) for some time after that, which doesn't match up. Still, it seems quite possible I broke it with theme changes.

(visible-control-characters-1 seems to have a few additional rare, flaky failures outside of those ranges.)