STR: 1. open http://kangax.github.io/compat-table/es6/ and wait until it's full loaded 2. scroll the page in different direction ER: the table header behaves responsively as expected as we scroll the page AR: there is a significant delay between the page is scrolled and the header is put in the right position I can reproduce this with both 47 and 48, with and without e10s enabled. Not reproducable in 45.

(In reply to Xidorn Quan [:xidorn] (UTC+8) from comment #0) > I can reproduce this with both 47 and 48, with and without e10s enabled. > reproducable in 45. Interesting. I can reproduce it in Nightly (49), but only with e10s (and therefore APZ) enabled.

The header doesn't actually use position:sticky - it appears to use the jQuery.floatThead plugin [1] to simulate position:sticky by changing the header's 'top' property in response to scroll events. They have logic that tries to use position:sticky instead, but only on Safari: if ($('thead').css('position') === "-webkit-sticky") { // Remove floatThead when native position:sticky is usable. // Currently, only Safari (which floatThead incidentally does not support), // and its -webkit-sticky prefix, works correctly with <thead>. $.fn.floatThead = function(){ return this }; } I believe that makes this an evangelism issue. [1] http://mkoryak.github.io/floatThead/

Xidorn, are you sure you can reproduce this with e10s disabled? My investigation is pointing to an issue that shouldn't be present with APZ disabled (an e10s disabled implies APZ disabled).

It seems I cannot reproduce this with e10s disabled, but I can reproduce it in a non-e10s window on Nightly.

(In reply to Xidorn Quan [:xidorn] (UTC+8) from comment #4) > It seems I cannot reproduce this with e10s disabled, but I can reproduce it > in a non-e10s window on Nightly. Thanks for checking! It looks like non-e10s windows still use APZ, so that's consistent.

As per comment 2, moving to Tech Evangelism.

Pinging Mike Taylor on this one.

(In reply to Botond Ballo [:botond] from comment #2) > They have logic that tries to use position:sticky instead, but only on > Safari: > // Currently, only Safari (which floatThead incidentally does not > support), > // and its -webkit-sticky prefix, works correctly with <thead>. Unfortunately a simple patch to that script isn't possible: we still don't support position:sticky with thead. (And I see kangax commented as much about his site in https://bugzilla.mozilla.org/show_bug.cgi?id=975644#c13).

This sounds similar to bug 1191995, although that describes a slightly different scrolling-related concern on the same page.

(In reply to Mike Taylor [:miketaylr] from comment #8) > (In reply to Botond Ballo [:botond] from comment #2) > > They have logic that tries to use position:sticky instead, but only on > > Safari: > > > // Currently, only Safari (which floatThead incidentally does not > > support), > > // and its -webkit-sticky prefix, works correctly with <thead>. > > Unfortunately a simple patch to that script isn't possible: we still don't > support position:sticky with thead. > > (And I see kangax commented as much about his site in > https://bugzilla.mozilla.org/show_bug.cgi?id=975644#c13). Oh, good catch - thanks for investigating! I'm going to move this bug back to Panning and Zooming until it becomes actionable by Evangelism again.