799775 - "-moz-order" needs to be inherited by ::-moz-table-outer in ua.css

Status: RESOLVED FIXED

(Core :: Layout, defect)

Description

[Daniel Holbert [:dholbert]]

ua.css needs to make table-outer frames inherit "-moz-order", alongside the existing rule to inherit "-moz-align-self".

This is necessary for table frames to obey "order" inside of a flexbox, for reordering purposes.

DETAILS:
The outermost frame on a table -- the frame that a flex container interacts with -- is the special "table-outer" frame,  which isn't directly stylable by content.  Its style-context's parent is the table element's style-context (though the parent/child relationship is reversed in the frame tree, paradoxically).

So, if we specify "order" on a table element and we want that to have a visible effect, we need the table-outer frame to inherit that "order" value, since that's the frame that the flex container ultimately interacts with.

Comment 1

[Daniel Holbert [:dholbert]]

Attachment: part 1: general mochitest for 'order'

First part: just a mochitest for general "order" behavior (reordering content within a flexbox).  (Wrote this as part of bug 666041, but didn't land it yet.)  The test reorders flex items with a variety of "order" values and compares their rendering against a reference case using WindowSnapshot.js.

I'm including this mochitest here (on this bug) because the *actual* regression-test here will depend on this mochitest. (I'll create a copy of the mochitest, using tables instead of blocks as flex items.)

Comment 2

[Daniel Holbert [:dholbert]]

Attachment: part 2: fix (tweaking ua.css) & table-specific mochitest

Here's the fix (adding a line to ua.css), and an "hg cp" of the other patch's mochitest into a table-specific one.

I've made this patch's table-specific mochitest finish early if it detects that the flexbox pref isn't already enabled (i.e. at startup).  I did this because because this test relies on the ua.css tweak, which only takes effect if the flexbox pref is enabled at startup (when ua.css is parsed).

Comment 3

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Comment on attachment 670107 [details] [diff] [review]
part 1: general mochitest for 'order'

Was including property_database.js a mistake?

r=dbaron (rubber-stamp)

Comment 4

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Maybe we could do something to always enable the parsing of the flexbox properties in the UA sheet so that we don't have the weird problems here?

Comment 5

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

Er, sorry, I meant that for bug 799647.

Comment 6

[Daniel Holbert [:dholbert]]

(In reply to David Baron [:dbaron] (recovering from illness; hopefully back Oct 8, but with backlog) from comment #3)
> Comment on attachment 670107 [details] [diff] [review]
> part 1: general mochitest for 'order'
> 
> Was including property_database.js a mistake?

You're right, property_database.js isn't needed -- thanks, removed.

Landed:
  https://hg.mozilla.org/integration/mozilla-inbound/rev/7e4cdcd41798
  https://hg.mozilla.org/integration/mozilla-inbound/rev/f858236f4e62

Comment 7

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/7e4cdcd41798
https://hg.mozilla.org/mozilla-central/rev/f858236f4e62