currentStatus.js for richtext only stores known fails. This way, if a known fail becomes a pass, you can just delete the line from currentStatus.js. For richtext2, it stores data for all tests, so a) it's much more cumbersome to update if you fix bugs; b) it's impossible to tell from the diff whether the change to currentStatus.js adds new expected failures or only changes/removes them; and c) it makes merge conflicts hard to impossible to resolve by hand. We only look at .selScore, .valScore, and .output. I'm not sure we really need to use .output -- the tests are very self-explanatory if you look at the webpage. I've never had to use it. Ehsan, would it be okay with you if instead of the TEST_RESULTS object, I make two objects called knownSelFailures and knownValFailures and drop the .output stuff? So it will give the test name on failure but no actual/expected values. Alternatively, I could maintain a third object with the actual/expected values for everything, if we want that too.

Diffstat: editor/libeditor/html/tests/browserscope/lib/richtext2/currentStatus.js | 26317 ++-------------------- editor/libeditor/html/tests/browserscope/test_richtext2.html | 71 +- 2 files changed, 2506 insertions(+), 23882 deletions(-) This would be great for me, because I wouldn't have to make all my richtext2-affecting patches depend on each other.

Comment on attachment 615663 [details] [diff] [review] Patch v1 Review of attachment 615663 [details] [diff] [review]: ----------------------------------------------------------------- I'd take any patches which make your life easier in a heartbeat! :-)