I found this while working on bug 1013219 and I'm breaking it out into a separate bug for clarity. The bug is that PCToLineNumber conditionally updates the line number, but unconditionally clears the column number. This means that in some cases the column number is incorrect.
Created attachment 8702683 [details] [diff] [review] rewrite PCToLineNumber The patch. Nick r+'d the basic idea in the other bug but I will re-request review since I've updated some tests.
Created attachment 8703016 [details] [diff] [review] rewrite PCToLineNumber Found one more regression from this.
Comment on attachment 8703016 [details] [diff] [review] rewrite PCToLineNumber Here's the PCToLineNumber patch separated out. It cleans up a few mysterious column number choices in the tests. You've already ok'd the original patch; not sure if you are able to review the test fixes as well?