Related info in *lreq documents:

• clreq
• jlreq
• klreq
• mlreq

Could also include Chinese newspaper example with both RTL for body text and LTR for captions…

@frivoal the above issue was raised with regard to the horizontal review on multicol, this is very much your area. Happy to do create and add an example, but I just wanted to ping you for your thoughts on what might be best to include before I draft something.

The issue with Chinese is more complicated. I grew up with papers that had RTL vertical text with RTL heads (whether horizontal or vertical). I used to work in a newspaper publisher (quite a while ago) and was told because there’s no horizontal RTL support in layout software, people actually typeset the RTL heads by typing the heads backwards.

We also now have even LTR vertical text (it’s been at least a decade now, if not longer) and obviously, there’s no layout software that can do this because non–Chinese speakers wouldn’t know there’s such a thing.

While it's nice to show that columns in a vertical text are a thing, this is a sufficiently complicated and subtle topic I'm not sure we should get too far into vertical text questions in the multicol spec. Something like the Mongolian example above is probably fine, as it's just multicol in vertical, but the earlier Japanese example involves page floats, and getting into mixed writing modes example with some text being vertical and some being horizontal, and dealing with whether the horizontal bits are LTR or RTL seem way out of scope for multicol. If we want a vertical-rl example as well, it should be something rather simple only involving multicol (possibly with spanners), to avoid overwhelming the reader.

I am a firm proponent of making sure all these things work and of enabling complex examples with mixed writing modes and mixed conditionality, but I don't think multicol is the right place to get into that.

I don’t think the Japanese example is complicated. There’s no concept of RTL Japanese in Unicode (it’s not done in Japanese these days in any case, as far as I understand; RTL horizontal text is only an issue in traditional Chinese and only in specialized applications), so ignoring the floated figure, all we have is a head followed by a vertical RTL div in four columns. I don’t think it’s any more complicated than the Mongolian example.

While i agree that the Japanese example isn't complicated, i also agree that the heading is not relevant here, and need not be in the example. The point of the example is solely to remind implementers that the flow of multi-column text will be different in vertically-set documents.

How about just using a schematic such as the following, where the middle represents Japanese/Chinese, and the right is Mongolian:

@r12a I like that, I think it demonstrates how it should work without introducing additional complexity that doesn't really belong in this spec.

@r12a I added images as requested, let me know if this is a satisfactory resolution. Thank you!

Added as example 12, in the section The Multi-Column Model: https://drafts.csswg.org/css-multicol/#the-multi-column-model

Thanks, the note and figure look good to me.

(The arrows in the figure seem to be at an angle. It may be a visual illusion, or may be intended. Not particularly important, but i was curious.)