A bunch of the APZ mochitests are gonk-only when they don't need to be any more. I've been working on getting those mochitests running on other platforms during my compiles - I'm not quite done but filing this bug to track the issue.

This patch adds the annotation in APZCTreeManager. It also changes the tree-building code in apz_test_utils to link together the actual paint structures rather than create new wrapper nodes. This is more foolproof (fixes a latent bug where a grandchild node whose parent has not yet been processed ends up with the parent forcibly made a child of the root), and allows extra properties to seamlessly ride along into the "tree". Review commit: https://reviewboard.mozilla.org/r/37135/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/37135/

Review commit: https://reviewboard.mozilla.org/r/37137/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/37137/

Review commit: https://reviewboard.mozilla.org/r/37139/diff/#index_header See other reviews: https://reviewboard.mozilla.org/r/37139/

Comment on attachment 8724741 [details] MozReview Request: Bug 1251608 - Add a root-content annotation to the APZ test data structure. r?botond https://reviewboard.mozilla.org/r/37135/#review34923 Thanks, this is quite a bit nicer! ::: gfx/layers/apz/test/mochitest/apz_test_utils.js:72 (Diff revision 1) > + if (!!apzcTree.isRootContent) { Is the "!!" necessary in an if condition?

Comment on attachment 8724742 [details] MozReview Request: Bug 1251608 - Fix and enable test_bug982141 for all platforms. r?botond https://reviewboard.mozilla.org/r/37137/#review34925 ::: gfx/layers/apz/test/mochitest/mochitest.ini (Diff revision 1) > -skip-if = toolkit != 'gonk' # bug 991198 Bug 991198 is about window.open not working on B2G desktop. Is that still relevant?

Comment on attachment 8724743 [details] MozReview Request: Bug 1251608 - Fix and enable test_bug1151663 for all platforms. r?botond https://reviewboard.mozilla.org/r/37139/#review34927 ::: gfx/layers/apz/test/mochitest/helper_bug1151663.html:65 (Diff revision 1) > - SimpleTest.is(apzcTree.children.length, 1, "expected a single root APZC"); > + // child APZC which is for the RCD (fennec case). I'd reword this to make it clear that in the e10s/B2G case, the single root APZC *is* the RCD.

Comment on attachment 8724743 [details] MozReview Request: Bug 1251608 - Fix and enable test_bug1151663 for all platforms. r?botond https://reviewboard.mozilla.org/r/37139/#review34929

https://reviewboard.mozilla.org/r/37135/#review34923 > Is the "!!" necessary in an if condition? I think technically no, but idiomatic Javascript uses !! when the thing you're testing might be undefined, which is the case here (isRootContent is either undefined, or "1" in JS). I'll add a comment to that effect as well to be more explicit.

https://reviewboard.mozilla.org/r/37137/#review34925 > Bug 991198 is about window.open not working on B2G desktop. Is that still relevant? I don't think B2G desktop is still a thing, it's been replaced by Mulet. That bug can probably be closed, and regardless it's not relevant to this ini file anymore.

https://reviewboard.mozilla.org/r/37139/#review34927 > I'd reword this to make it clear that in the e10s/B2G case, the single root APZC *is* the RCD. Done.

https://hg.mozilla.org/integration/mozilla-inbound/rev/90e4525b2989 https://hg.mozilla.org/integration/mozilla-inbound/rev/b592bfa4541d https://hg.mozilla.org/integration/mozilla-inbound/rev/954f09e238ca

Oh, since I put these patches up we started running M-e10s on OS X as well. Hopefully they don't start failing there (they worked locally but I don't think I tested them on try...). If there's issues I can back out.

https://hg.mozilla.org/mozilla-central/rev/90e4525b2989 https://hg.mozilla.org/mozilla-central/rev/b592bfa4541d https://hg.mozilla.org/mozilla-central/rev/954f09e238ca