Spun off from other work. This causes horrible logspew.

Fix some borked indentation.

Comment on attachment 8380310 [details] [diff] [review] ContentProvider tests don't clean up their cursors. v2 Review of attachment 8380310 [details] [diff] [review]: ----------------------------------------------------------------- Not running tests locally for the loss! ::: mobile/android/base/tests/testBrowserProvider.java @@ +1836,5 @@ > ensureOnlyChangeNotifiedStartsWith(BrowserContract.History.CONTENT_URI, "bulkInsert"); > } > } > + > + void assertIsAndClose(Cursor c, int expectedCount, String message) { |assertCountIsAndClose|?

Pushed with name change, and added docstring and `private`: https://hg.mozilla.org/integration/fx-team/rev/49a1923db9be

Follow-up, 'cos the favicon check that I re-enabled fails on Tinderbox but not on my device. https://hg.mozilla.org/integration/fx-team/rev/1f6450d61d3f

https://hg.mozilla.org/mozilla-central/rev/49a1923db9be https://hg.mozilla.org/mozilla-central/rev/1f6450d61d3f

Backed out for too frequent robocop failures (bug 975969 and more). https://hg.mozilla.org/mozilla-central/rev/1a219f500336 https://hg.mozilla.org/mozilla-central/rev/1422dfcd7fd8

Good guy rnewman leaves a note in the Sheriff's etherpad, gets backed out

Note that the test previously did not check what it was supposed to check. So this push: https://tbpl.mozilla.org/?tree=Try&rev=e4e2e255d5ab just logs any irregularities, rather than actually asserting them. If it's something obvious, I'll fix; otherwise, I'll just reland it without the checks that fail.

Relanded with the failing checks commented out. https://hg.mozilla.org/integration/fx-team/rev/9122f3be8c3a

Yeah, when your note is in effect saying "even if I landed 50% orange I don't want to be backed out" you don't want to put it in that etherpad, you want to put it... um, in... uh..., I think the only place you could put that note would be in the second line of the commit message for the cset that completely disables the test that you were otherwise going to make 50% orange. BTW, so far on the reland, https://tbpl.mozilla.org/php/getParsedLog.php?id=35167881&tree=Fx-Team

Huh, build didn't fail locally. Also, this would be way less painful if TBPL would load.

Apparently CursorDumper is only available at build time, not at run time. https://hg.mozilla.org/integration/fx-team/rev/efa293b98893

Relanded after three green Try rc1s, without the foolish expectation that our core database code works. https://hg.mozilla.org/integration/fx-team/rev/4881a4e25a3e

(In reply to Richard Newman [:rnewman] from comment #9) > Note that the test previously did not check what it was supposed to check. > So this push: > > https://tbpl.mozilla.org/?tree=Try&rev=e4e2e255d5ab > > just logs any irregularities, rather than actually asserting them. If it's > something obvious, I'll fix; otherwise, I'll just reland it without the > checks that fail. BTW, this is filed as bug 949048.