[css-display][css-flexbox] Blockification of layout-internal type

[Loirooriol]

CSS Display defines blockification as such:

blockification [...] sets the box’s outer display type, if it is not none or contents, to block [...]

Some trivial examples:

• A flex item with display: block (i.e. block flow) remains block flow (and establishes a BFC).
• A flex item with display: inline (i.e. inline flow) becomes block flow (and establishes a BFC).
• A flex item with display: inline-block (i.e. inline flow-root) becomes block flow-root.
• A flex item with display: flow-root (i.e. block flow-root) remains block flow-root.
• A flex item with display: table-cell (i.e. table-cell flow-root) becomes block flow-root.
  Note: Edge generates anonymous table parents instead, according to an old Flexbox spec.
• A flex item with display: table is especially defined in CSS Flexbox.

But now the interesting part:

• A flex item with display: table-row (i.e. table-row table-row) becomes block table-row.

But this combination is not defined! How is it supposed to behave?

Firefox and Chrome seem to treat it like display: block. Edge still generates an anonymous parent table. https://jsfiddle.net/sn6y4y0m/1/

I think CSS Display should say that, when a box is blockified, if its inner display type is layout-internal, then it becomes flow or flow-root.

[Loirooriol]

And similar for inlinifications. That's only a theoretical problem because as far as I know they only happen in ruby, and unlike flexbox, ruby says they happen after anonymous-box fixup. But CSS Display should give a proper definition.

[tabatkins]

I think the right thing to add is:

If the box originally had a [=layout-internal=] display type,
then its [=inner display type=] is also set to ''flow''.

wdyt, @fantasai?

[tabatkins]

@fantasai approved this in person 👍