The following layout tests are failing in TOT due to firstCharacterForRange(0, 0) returning a 0x0 rect: editing/input/firstrectforcharacterrange-plain editing/input/firstrectforcharacterrange-styled fast/text/attributed-substring-from-range It looks like a regression from fixing bug 6246.
Created attachment 5369 [details] band-aid fix Looks like an indirect result of issues described in bug 5610: the range returned for (0, 0) in these tests is (#text, 1)-(#text,1) rather than (#text, 0)-(#text,0) - and the latter or equivalent don't give correct results in firstRectForDOMRange (which may be a separate issue). Going to cleanup this further as part of the investigation on 5610...
r=me, looks like an ok fix for now if it gets the tests passing and doesn't break anything.