Regression? Whitespace-only differences in test output across platforms Jeffrey Pfau noticed there were some whitespace-only differences between output on some of the testing bots for a few tests including accessibility/aria-readonly. We don't think the bug is specific to the test, but do want to track this in case it represents another regression. My only hypothesis is that the whitespace-flattening algorithms may have regressed.
<rdar://problem/14447032>
Are there any new features involved, that may be disabled on bots?