WHATWG on 2023-08-12

05:46	<Nicholas Shanks>	Hello. I guess this is the new USENET/IRC.
05:48	<Nicholas Shanks>	I recently discovered that there is no default value for the @scope attribute of the TH element. I always thought it was `row` for children of TBODY and `col` otherwise. Is that not how browsers behave anyway?
05:51	<Nicholas Shanks>	If confirmed, would it be wise to codify that behaviour in the standard? (and, is Ian Hickson still the editor?)
06:18	<Nicholas Shanks>	Hmm, the current developer presentation of the WHATWG edition of the standard differs from waht is published on MDN: https://developer.mozilla.org/en-US/docs/Web/HTML/Element/th#scope says "browsers automatically select the set of cells to which the header cell applies" but does not mention the `auto` state, leaving me to think it was up to browsers to create some heuristic of their own. This page https://html.spec.whatwg.org/dev/tables.html#attr-th-scope gives quite a verbose description of the expected behaviour, along with diagrams, but does not specify that succinctly in the description of the `auto` state.
06:23	<Nicholas Shanks>	The description explicity says both `auto` and `rowgroup` do not apply to cells preceding the TH with that scope in the current row (for TH children of TBODY) or rowgroup. It says nothing about if/how the `auto`, `col` and `colgroup` states apply headers to preceding cells.
06:29	<Nicholas Shanks>	Also, the described behaviour of an explicit `rowgroup` value, to not apply to cells in preceding columns, differs from what I would intuitivly have expected. In the diagram, `<th> Cats` does not apply to `<td> 93`. Is that really how browsers and accessibility agents behave? In other words, `rowgroup` does not mean "rowgroup", it means "cells in this and following columns within this rowgroup"
06:38	<Nicholas Shanks>	More discrepencies: The row group state means the header cell applies to all the remaining cells in the row group. This wording is incorrect according to the Cats/Legs diagram. The behaviour illustrated would correspond to: "The row group state means the header cell applies to all cells in the current and subsequent columns, within the row group."
06:44	<Nicholas Shanks>	If "remaining cells" was defined to mean cells in remaining columns, would that include cells that span across from a preceding column?
07:19	<Nicholas Shanks>	`scope` was added between September and November 1997: https://www.w3.org/TR/PR-html40-971107/struct/tables.html#adef-scope `row`: The current cell provides header information for the rest of the row that contains it. etc. So, "the rest of" is clear for row. `rowgroup` is similarly defined, so it is not clear if it includes preceding columns in subsequent rows (as might be interpreted)
07:24	<Nicholas Shanks>	Also the WHATWG text "The remaining headers apply just to the cells to the right of them." should be amended to use logical directionality for RTL tables.
07:32	<Nicholas Shanks>	So, my findings are that the definition of `row` and `col` is consistent with the behaviour of `auto` when the TH is a child of a TBODY, or not, respectivly, though this behaviour is not explicitly defined for auto.
07:36	<Nicholas Shanks>	In other words, the definition of `auto` could be changed from The auto state makes the header cell apply to a set of cells selected based on context. to e.g. "The auto state behaves as if it were set to row if `tbody > tr > th:not([scope])`, and `col` otherwise." I think this would be a more useful definition.