[css-contain] editorial: Paint Containment description is incomprehensible

[MatsPalmgren]

ISSUE 1:
The Paint Containment section says (1st bullet):

The contents of the element including both the paint of the descendants and their geometry ...

The term "the paint of the descendants" is unfamiliar to me and I can't find any definition of it in any CSS spec.
The term "geometry" is unfamiliar to me and I can't find any definition of it in any CSS spec.
(I really have no clue what you mean by either of those in this context.)

Please provide precise definitions of what you mean by these terms, or preferably, use well-known CSS terms that are defined in other specifications such as: scrollable overflow, ink overflow, content-box size, border-box size etc... (please also linkify at least the first use of such terms for extra clarity). Thanks!

ISSUE 2:
The last sentence under the 1st bullet reads: "This is as if overflow: visible was changed to overflow: clip at used value.". What does "this" refer to in that sentence? Is it the whole paragraph, or just the preceding sentence, or perhaps just the first sentence?
The only interpretation that makes sense to me is that it refers to the first sentence, since a scrolling mechanism is never created for overflow:clip. If so, then it seems better to insert this sentence directly after the first sentence. Or perhaps just re-write the first sentence with "overflow:visible behaves as overflow:clip" included in some way. (Again, it's better to use a familiar term like "behaves as" than trying to mint your own version of it.)

[frivoal]

Issue 1:

I think we can replace "the paint of the descendants" with "any ink or scrollable overflow" without altering the effect. I think the original intent of "the paint" was to talk about things that would be ink or scrollable overflow if they overflowed, but in this case we only care about what happens to that when it does overflow, so we can talk about the overflow using established terminology instead.

• I am less sure about what "and their geometry" was supposed to be about. @tabatkins ? Maybe it is supposed to be about was outlines (and possible other things like outlines): since outlines are pretty flexible about what they can be around, the "geometry" of anything that pokes out of the containing box might be taken into account, but this was trying to say that it must not be. If that's that, maybe it should be moved into a separate and clearer bullet point. If it's something else, what?

Issue 2:
It refers to the whole paragraph: it only behaves as overflow: clip if overflow was otherwise visible. If it was some other value, it says unchanged. Hence the "does not include the creation of any mechanism… nor does it inhibit…".

How about rephrasing the note into

The behavior is described in this paragraph is equivalent to changing overflow:visible into overflow:clip at used value time, while leaving other values of overflow unchanged.

[tabatkins]

I am less sure about what "and their geometry" was supposed to be about.

I'm referring to layout overflow.

[tabatkins]

Ugh, scrollable overflow, whatever. Overflow that causes scrollbars to happen.

[frivoal]

Issue1:

The contents of the element including both the paint of the descendants and their geometry

becomes

The contents of the element including any ink or scrollable overflow

---

Issue 2:

This is as if overflow: visible was changed to overflow: clip at used value.

becomes

The behavior is described in this paragraph is equivalent to changing overflow:visible into overflow:clip at used value time, while leaving other values of overflow unchanged.

---

Ok for everyone?

[MatsPalmgren]

Issue 1: @tabatkins said "scrollable overflow", you said "ink or scrollable overflow". It seems you disagree?
If you want visible to behave as clip then it should be the latter since CSS Overflow says that overflow-x/y also clips ink overflow.

Issue 2: sounds good to me if you change all instances of "overflow" to "overflow-x/y" in that sentence, so that it's clear also when just one is visible.

[tabatkins]

@tabatkins said "scrollable overflow", you said "ink or scrollable overflow". It seems you disagree?

No, I was explaining "geometry" as scrollable overflow. Florian is replacing the whole "paint and geometry" reference, which is indeed both ink and scrollable overflow. No disagreement.

[css-meeting-bot]

The CSS Working Group just discussed editorial: Paint Containment description is incomprehensible, and agreed to the following:

• RESOLVED: Accept the comment at the end of https://github.com/w3c/csswg-drafts/issues/4946#issuecomment-614501289

The full IRC log of that discussion

<dael> Topic: editorial: Paint Containment description is incomprehensible
<dael> github: https://github.com//issues/4946
<miriam> thanks! :)
<dael> florian: Matts has found there's places we're vague and non-standard. First is we mention things that must be clipped include content and paint and geometry of desc. Not defined terms. I think it was ink and scrollable overflow
<dael> florian: Should replace the phrasing
<dael> TabAtkins: Yeah
<faceless2> @orial awful for me too
<dael> florian: Second part is in a note we say as if overflow-visible was changed to overflow clip as used value. Prop we replace with [reads]
<dael> florian: I'll tweak to be explicit about overflow-x and -y in case it's different values
<dael> Rossen___: Sounds great
<dael> smfr: Is scrollable overflow defined?
<dael> florian: Yes
<dael> smfr: descendants you need to be clear paint order? containing block?
<fantasai> smfr, definitions of overflow: https://www.w3.org/TR/css-overflow-3/#overflow-concepts
<dael> smfr: I mean z order descendants. Stacking context
<dael> florian: If we mean all do we need to disambig?
<chris> +1 to smfr say which tree they are descendants in
<dael> smfr: All is unclear. dom node? A child can break out if position abs. I may not paint as desc if it has z-index
<dael> florian: I think we're talking about all starting from dom. Weither positioned or painted they're part of all
<dael> smfr: THan have to create containing for positioned and stacking
<dael> florian: THink it does
<fantasai> https://drafts.csswg.org/css-contain-1/#containment-paint
<dael> smfr: And containing block for fixed?
<dael> dbaron: Does for fixed and abs and stacking contexts. Bullets 2 & 3
<dbaron> in https://drafts.csswg.org/css-contain/#containment-paint
<dael> Rossen___: Any other questions or opinions or good to resolve?
<dael> florian: One question. Correct phase to say all descendants? All DOM?
<dael> fantasai: All ambig reasolves to same thing doesn't matter.
<dael> florian: We have later bullets to say if it's all it works.
<dael> fantasai: What's interp that doesn't mean all? ambig can't resolve to anything that's not what you meant so it doesn't matter
<dael> smfr: It's fine
<Rossen___> https://github.com//issues/4946#issuecomment-614501289
<dael> Rossen___: Objections to proposal here ^
<dael> florian: Prop: Accept the comment at the end of https://github.com//issues/4946#issuecomment-614501289
<dael> RESOLVED: Accept the comment at the end of https://github.com//issues/4946#issuecomment-614501289
<dael> fantasai: Want to resolve to update recommendation?
<dael> Rossen___: This is WD
<dael> florian: L1 is a rec
<dael> florian: L2 is a WD, but L1 is rec and phrase is in both
<dael> Rossen___: oooh
<dael> chris: Should be an erratum. Are there any others we can roll in to a publish?
<dael> florian: I need to check
<dael> florian: Let's move on with agenda and I'll get back about if there are more in the queue
<AmeliaBR> No errata for contain 1: https://www.w3.org/Style/2019/REC-css-contain-1-20191121-errata.html