[mediaqueries-4] Drop overflow-block:optional-paged

[frivoal]

The overflow-block media feature describes the behavior of the device when content overflows the initial containing block in the block axis. The none, scroll, and paged values correspond to well known behaviors, but optional-paged is more theoretical/speculative, and doesn't really correspond to anything that currently exists.

It was meant to match things like the (presto-based) Opera <=12 presentation mode. This was a user-activated mode, fundamentally similar to scroll (overflow triggers scrollbars; the ICB is the viewport, not the page area; and the whole css-page model doesn't apply), but in addition, it allowed the author to insert forced fragmentation breaks, leading to something akin to a paginated behavior. This was meant for making things like slide decks.

While I think experimenting with that sort of things is great, no current user agent behaves like that, so no current user agent will match that particular value. This makes it not useful.

Further, if someone were to experiment again within that space, we cannot be sure that the behavior they'd land on would be the same as the one opera had and that is currently used in the spec as the definition of optional-paged, which would make the value misleading.

If someone ever ships something of that nature, we should definitely look into making sure that whatever behavior they end up with can be covered by media queries, but until that happens, I think we should drop this speculative value.

[fantasai]

I think it makes more sense to mark it at-risk, and drop it when you go to REC if there's no implementation.

[frivoal]

I'd be OK with at-risk, but I'd actually prefer removing it: the problem is not merely lacking implementations of this media query, but rather that it is unclear to me whether that media query is specified right: the class of UA it purports to describe doesn't currently exist, and if something similar were to be created, we don't know that they would actually behave the way the spec claims they would.

[fantasai]

It did exist in the past, though, so it's not merely theoretical. Also, removing it from the specs means any UA that has such a combination mode would almost certainly just assume there's no value that matches their behavior and will pick one (or both) of the existing values to match. Leaving it in the spec is more likely to result in them filing an issue about the description being slightly off.

[frivoal]

If you want to leave it in at-risk, I'd suggest switching it from counting as "paged media" to counting is "continuous media" for the purpose of defining these two terms, as the historical opera behavior had more in common with the former than the later in terms of definition of the ICH, responding the the viewport meta tag, not responding to @page, etc.

My preference towards removing it is in part because I don't like answering these questions in the abstract (what would the viewport units have matched in opera 12's presentation mode had it had them), but if we're leaving it in, that historical behavior was closer to continuous than to paged.

[css-meeting-bot]

The CSS Working Group just discussed [mediaqueries-4] Drop overflow-block:optional-paged, and agreed to the following:

• RESOLVED: Drop the value with a note explaining why

The full IRC log of that discussion

<dael> Topic: [mediaqueries-4] Drop overflow-block:optional-paged
<dael> github: https://github.com//issues/5287
<dael> florian: We have the overflow:block MQ which lets you ask if you're paginated or scrolling document when you overflow
<dael> florian: One value, paged, is optional. Hybrid and last found in Opera 12. Looked normal but if you put a forced break you get a new page. If user set browser for presentation mode. It's like screen but you can get pages. That's optional paged.
<dael> florian: Opera 12 no longer ships and no other UA does this. I don't expect it impl any time soon.
<dael> florian: Also, I'm not sure next time someone experiments with that kind of behavior that they'd do it like Opera so I'm not sure spec should define how impl behaves
<dael> florian: In favor of dropping this. fantasai argued mark at-risk which is normal for spec but not impl. Since I think ti's not clear that mode wouldn't be used as spec prefer drop
<dael> tab: Agree with florian
<dael> AmeliaBR: No reasonable expectation of any impl to match this between CR and REC?
<tantek> q+ to ask since it has shipped, there may be content, should we mark it obsolete?
<dael> florian: We wouldn't. Interesting reason to leave it in is if someone wants to experiment to something like that they wouldn't map to the other values. Could be a note saying if you're doing none of these come talk
<dael> tantek: Could leave with either option. Conern from content side where if it did exist there might be content out there using it. Could we consider 3rd option to mark as obsolete to say it was exposed to web
<dael> fantasai: MQ wasn't
<dael> florian: Browser that behaved like this existed, but didn't have MQ
<dael> florian: Browser shipped the behavior but didn't have the value
<dael> tantek: How did they ship?
<dael> florian: If user pressed F11 they would get forced pages, but no MQ to get that
<dael> fantasai: Responded to presentaiton media-type so that's how you could get it. It was before MQ4 existed
<dael> tantek: Obsolute the presentation-type already?
<dael> florian: Ye
<dael> tantek: Okay dropping. Also okay in a draft as at-risk and it will be dropped in next draft. Gives public a chance.
<dael> stearns: We have a good signal from impl that none are looking into this. Prefer to drop though happy to have a note we're dropping without prejudice. No implementations now but we're open to experimentations
<dael> florian: Put note in section so if UA have interesting another mode you should talk to us since we had an interest.
<dael> tantek: But looking for impl interest
<dael> stearns: Prop: Drop the value with a note explaining why
<tantek> +1
<dael> RESOLVED: Drop the value with a note explaining why
<dael> fantasai: New publication?
<dael> florian: I have DoC and changes section and tests. Before resolving I do want to ask for another else to be at-risked.
<dael> stearns: So that's got to go into next week's agenda.
<dael> stearns: Thanks all