Unclear use of the term "descendant"

[WilcoFiers]

There are a number of places where the word "descendant" is used. It is unclear what tree this is referring to. There are different trees involved in the user agent, such as the light DOM tree, shadow DOM trees, the flat tree, and the accessibility tree. In some uses of the word "descendant" the intended purpose seems to be flat tree, in others a document fragment, and yet others the accessibility tree.

This problem is new in ARIA 1.2 because it is written for DOM 4, rather than DOM 3 which was used in ARIA 1.1..

[css-meeting-bot]

The ARIA Working Group just discussed Unclear use of the term "descendant".

The full IRC log of that discussion

<carmacleod> github: https://github.com//issues/1150
<carmacleod> jamesn: scott owns - he is the right person to take this on
<carmacleod> jamesn: does need to go in 1.2, because DOM 4 and not DOM 3 - our references have changed
<carmacleod> sina: I agree that it's editorial

[jnurthen]

@scottaohara any updates on this one?

[cookiecrook]

Also see #1161 which refers to owned elements in the context of descendants.

[jongund]

@jnurthen

You may want to wait until PR 1162 is merged, since there will be more references to descendant in the document before reviewing the document for this issue.

[jnurthen]

Discussed in WG call on 11/12/2020

**JN:** so the first one of these is 1150, and this is unclear use of the term descendant … I think at this stage going through and doing a PR on this isn't necessarily the right way of going about it. I think we need to go through and work out all the different uses of descendant that we have … doing an informal search of the spec, we have 4 different uses. There's DOM descendant, descendant element (not sure if the same as DOM descendant), and then we have other uses of descendant, many of which are in activedescendant and keyboard handling sections, and then we have random other things … my proposal on this is that someone takes a look and tries to categorize the usages of the word descendant into various things and then we can potentially come up with a different word for some of them and a definition, or a definition for what descendant means in each of these ways … e.g. "element descendant" and "DOM descendant", and define each of those … so anyone want to voluteer for this?

SH: I wouldn't mind doing some of that

MK: do they have to be fixed before CR?

JN: no, only if they (the ACT) say they have to be
… so I'm not proposing fixing this in 1.2, I'm proposing as soon in 1.3 as we can, and getting that in the working draft as soon as we can

carmacleod: Also #1033 and #1161

MK: so you're not planning of taking the results of this analysis and trying to get something changed in 1.2 after it's gone into CR

JN: we are going to try not to do that, yes. We don't want to put anything new into 1.2, aside from editorial changes, and this is not editorial
… I put a 1.3 blocking label on this to make it clear what we are doing
… realistically if we recycle 1.2 and have to do some of the process again, it's not going to be any quicker
… if you want to work with me, I'm happy to do it

SH: sounds good, I'll start with writing up what's currently in the spec and leaving recommendations off for now

[smhigley]

I summarized all current uses of "descendant" in the spec in a gist

@jnurthen let me know if this is more or less what you were looking for :D

[pkra]

Assigning myself to see if I can help get it unstuck - no promises.

[spectranaut]

Discussed in F2F meeting yesterday: https://www.w3.org/2023/05/04-aria-minutes#t01

Decision to add "accessibility descendant" as an important term, to mirror "accessibility child" and "accessibility parent".

Then we can use "accessibility descendant" or "DOM descendant" as appropriate in the spec.