[css-rhythm-1] Inherit block-step-size

[fantasai]

Currently block-step-size functions both to set the step size and also to set whether the element honors the step size. It would be nice if it just set the step size and was inheritable: that way the author could set the step size once per document (or per section of the document) and just turn the stepping on or off with a different propert (such as block-step-align).

[jensimmons]

I believe this issue is asking this question:

Should block-step: 1lh make block stepping start happening, lining up to the nearest 1lh?

Or should block-step: 1lh only set the measurement, and another property turn on the behavior? Perhaps block-step-toggle: on? Ok, not that, but something that means that.

I don't believe we need another toggle. I don't think that block-step-align is the right mechanism.

margin-block: 1lh “turns on” margins.
align-items cascades through even when it has no effect.

Which makes me feel like block-step: 1lh should turn on stepping. And block-step-align should cascade through, even when it has no effect. Not the reverse.

[jensimmons]

For context, there are four longhands:

• block-step-size none | length (sets the increment of stepping & currently turns it on)
• block-step-insert margin-box | padding-box | content-box (says where the extra space goes)
• block-step-align auto | center | start | end (sets alignment once there is extra space)
• block-step-round up | down | nearest (sets whether round up/down/nearest when stepping)
  plus the block-step shorthand.

[jensimmons]

Now that block-step-size no longer creates a new formatting context (#11325), block-step-size: none will go away, and that way to "turn it off" will be block-step-size: 0;

[css-meeting-bot]

The CSS Working Group just discussed [css-rhythm-1] Inherit block-step-size, and agreed to the following:

• RESOLVED: Close no change

The full IRC log of that discussion

<bramus> jensimmons: block-step is a shorthand for 4 longhands
<bramus> … see comment for details
<bramus> … there is an idea to have 5 longhands
<bramus> … to have block-step-size not set the increment
<bramus> … so they can turn it on/off for specific situations
<bramus> … discussed with elika that it reminds me of margin
<bramus> … you dont set it in one location and then turn it on
<bramus> … suggestion to close no change
<bramus> … do not need a separate property to turn it on or off
<flackr> I feel like usually you'd use a variable to pass it down if you wanted to do this
<bramus> astearns: trying to think of th euse case where you want to keep an inherited block-step length and turn that on or off dependin gon the context
<bramus> … are you, elika, ok with closing no change?
<bramus> fantasai: no strong opinion
<bramus> iank_: might want to split it off if ??? child
<bramus> … to see if the thing as a whole
<bramus> fantasai: to set it on the figure an dnot the figcaption
<bramus> iank_: or an article
<bramus> astearns: there you turn it off for the children
<bramus> … usecase needs to be that you set it on a parent but dont want a descendant to use it but do want its children to use
<bramus> … extra footgun of setting block-step-size and having it not happen seems more like a general case that we should be solving for
<bramus> fantasai: for comparison, now that we dropped the new BFC requirement, we could drop the none value in block-step-size with an initial value of 0
<bramus> … when you increase you get stepping
<bramus> … alt model is to be like text-box-trim where one prop sets the setting and another that turns it on and off
<bramus> … alt model where block-step-size inherits and you enable it on the boxes that need it
<bramus> … not convinced that that extra indireciotn is nessary
<bramus> … but have done it before
<bramus> s/nessary/necessary
<flackr> q+
<bramus> … cases where we did this are cases where you set a thing on a parent and cascade it individually
<astearns> ack flackr
<bramus> flackr: officially this seems like a use case for variables … to pass down values that are not applied
<bramus> … authors can use a variable if they want access to the value
<bramus> PROPOSED RESOLUTION: Close no change
<bramus> RESOLVED: Close no change

[fantasai]

@jensimmons were looking at some examples and realized that, actually, we do need block-step-size to inherit. Otherwise it's not possible to use a step size that's based on the font-size or line-height of an ancestor.

For example, suppose we have an <article>. We want headings inside it to use a step-size of 1lh of the article's line-height, not the heading's.

To make this work reasonably, we'll need the 1lh to compute on the <article> (and therefore inherit down as an absolute size).

So the way this works would be:

• block-step-size: <length> inherits, but has no effect
• block-step-align: none | auto | start | center | end causes block-step-size to have an effect on the element to which it's applies.

Re-opening for discussion.