We should test these from the timeline actor, not from the front. Bonus points if they are easily copy/pastable for new markers

Moved e10s process communication into a helper file so it can be shared on both browser/ and toolkit/ tests, and moved marker tests to toolkit/* https://treeherder.mozilla.org/#/jobs?repo=try&revision=f0ddc12373e5

Comment on attachment 8649963 [details] [diff] [review] 1165994-marker-tests.patch Review of attachment 8649963 [details] [diff] [review]: ----------------------------------------------------------------- Please maintain history. Use hg move to move files, qrefresh, then apply any changes on top if necessary. I know you hate hg. I'm not doing this to make you love it.

Most of the marker tests are completely different (and more consistent now) -- but the moves are much more cleaner now

Also some intermittent's on markers have negative time values (both start and end) -- this seems like it makes sense, and we just filter them out, so maybe we should get rid of that assertion? https://treeherder.mozilla.org/#/jobs?repo=try&revision=f0ddc12373e5 toolkit/devtools/server/tests/browser/browser_markers-gc.js | All markers have a start time between the valid range.

Backed out for M-e10s(dt2) failure on Linux: https://hg.mozilla.org/integration/fx-team/rev/19c7a4a7cefe

Issue with browser.ini, didn't maintain the skipping on linux (always forget the directives are below the lines they apply to), fixed up