If new-run-webkit-tests does not run a test, it should not report anything about that test; in particular, reporting that it unexpectedly passed when it wasn't run at all is very confusing.
We used to have a bug where we would log 'unexpected pass' for expected image failures (bug 36771), but that was fixed years ago.
When I run NRWT now, I don't see any "unexpected pass" messages for tests expected to fail with IMAGE failures.
However, I'm not sure I understand what you mean by "wasn't run at all"; you mean if a test foo.html is skipped, you're seeing unexpected pass for foo.html ?
If you can still reproduce this, can you tell me what command line you're running with?
Created attachment 148211 [details]
Here's the issue. I did run-webkit-tests --debug (i.e. NO pixel testing) but the results page said that sometests were expected to fail, but passed.
That's nonsensical, because only IMAGE failures were expected, but I'm not testing pixel results.
I see, I didn't realize we were talking about results.html ...
this is happening because the html/js page doesn't know if pixel testing was disabled or not. This should be easy to fix.
Created attachment 148221 [details]
Committed r120736: <http://trac.webkit.org/changeset/120736>