mozilla :: #layout

20 Mar 2017
10:55emiliodbaron: ping? I'm refactoring UndisplayedMap to contain a LinkedList instead of the manual UndisplayedNode list, and I planed to drop this bit, which seems fixed now (http://searchfox.org/mozilla-central/rev/c9f5cc6b4593d461e87a6221dc0286b3859fd515/layout/base/nsFrameManager.cpp#754)
10:55emiliodbaron: heycam told me to ask you if you think it was fine. In particular, it allows insertions in the undisplayed map to be O(1)
17:17hunboyfantasai, ping. Any plan in the future with "equal gutter" @ grid spec 1 or 2 ? https://jsfiddle.net/utasir/20yy425r/12/ perhaps "grid-row-gap: auto" value can be the "equal" stuff, or something similar ? This issue occurs when we define justify-content: space-between. And more difficult issue with space-around.
20:15dbaronemilio: I think if you assume it's fixed, you should make it a MOZ_ASSERT so that we find out if you're wrong
20:16dbaronemilio: but I'm not particularly knowledgable about whether it would be fixed.
20:16dbaronemilio: (I'd also maybe worry about imagemap areas)
20:16emiliodbaron: yeah, I added a proper assertion walking up the list in debug builds. It seems like no test in try hits that assertion.
20:17dbaronemilio: ok. (Keep an eye out for fuzzer bugs, though.)
20:17emiliodbaron: ok, thanks for the reply! :)
21:32tnikkeldholbert: did you figure out the scrollbar thing? you've got me curious now
21:32dholberttnikkel: nope, got a backlog of reviews over the weekend
21:32dholbertno coding today I think
21:59dholberttnikkel: while I've got you though -- I'm looking at some nsView stuff, and I'm wondering if this "operator delete" decl makes any sense (isn't it just the same as the default compiler-provided one?): http://searchfox.org/mozilla-central/source/view/nsView.h#65-66
22:00dholberttnikkel: (searchfox says that operator was introduced in a patch you r+'d in https://bugzilla.mozilla.org/show_bug.cgi?id=1232696 , "part 1" there)
22:00firebotBug 1232696 FIXED, jseward@acm.org Remove NS_DECL_AND_IMPL_ZEROING_OPERATOR_NEW as it causes segfaulting for GCC 6 builds
22:06tnikkeldholbert: hmmm
22:06tnikkeldholbert: that seems to come from the old definition of the NS_DECL_AND_IMPL_ZEROING_OPERATOR_NEW macro
22:06tnikkeli wonder if it was needed because we call operator delete directly? http://searchfox.org/mozilla-central/source/view/nsView.cpp#188
22:07dholberttnikkel: ah, perhaps! I wonder why we even do that
22:07dholbertoh, for poisoning I guess
22:07tnikkelyeah, to try to get sooner crashes
22:07tnikkeland to prevent UAF
22:08dholbertright. I misremembered that we only did poisoning for arena-allocated stuff
22:08dholbertbut I guess we do it for heap-allocated views too
22:08dholberttnikkel: thanks
22:08tnikkeldholbert: yeah, we specifically added it for views (even though it may not be useful) to try to get better stacks for some crashes
22:15dholbertgotcha
22:57emiliodholbert: ping?
22:58emiliodholbert: unping, need to run, will check with you other day :)
21 Mar 2017
No messages
   
Last message: 124 days ago