(In reply to Johann Hofmann [:johannh] from comment #27) > Please take your time with the reviews, I'm not hoping to land before 66 merges to beta. This feels like it would need a good beta cycle to confirm it doesn't impact perf and to iron out possible issues. Can you do a trypush with talos and raptor tests (at least 5 runs of each) to check for perf issues pre-emptively, if we expect there might be some? Did you consider an alternative solution, like refreshing the list when the tab gets a pageshow event (ie. because the user switched to a different tab and then back)? That seems like it'd require less infrastructure and have less perf impact. If we need to use this mechanism, can we add/remove the progress listeners based on when the dialog is actually open, instead of doing it on startup? As it is, the perf impact even just on startup seems like it'd likely not be acceptable (even if it probably won't end up showing up on try because it's so small on the really fast hardware we run the tests on, and the noise is substantial).
Bug 1460768 Comment 28 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
(In reply to Johann Hofmann [:johannh] from comment #27) > Please take your time with the reviews, I'm not hoping to land before 66 merges to beta. This feels like it would need a good beta cycle to confirm it doesn't impact perf and to iron out possible issues. Can you do a trypush with talos and raptor tests (at least 5 runs of each) to check for perf issues pre-emptively, if we expect there might be some? Did you consider an alternative solution, like refreshing the list when the tab gets a pageshow event (ie. because the user switched to a different tab and then back)? That seems like it'd require less infrastructure and have less perf impact. If we need to use this mechanism, can we add/remove the cache/cookie/whatever listeners based on when the dialog is actually open, instead of doing it on startup? As it is, the perf impact even just on startup seems like it'd likely not be acceptable (even if it probably won't end up showing up on try because it's so small on the really fast hardware we run the tests on, and the noise is substantial).