[css-inline] Specify fallback baselines estimations for text?

[yiyix]

In the CSS spec, hanging baseline is defined as the following Corresponds to hanging baseline from which characters in Tibetan and similar unicameral scripts with a strong but not absolute top edge seem to “hang”. Assigned to hang in OpenType and to bsln value 3 in TrueType AAT.

It didn't describe what's the fallback value for hanging baseline if "hang" is missing in OpenTyoe table. The definition is not entirely clear as well, it only says a strong but not absolute top edge. So it allows browsers to have their own interpretation of the baseline value if "hang" is missing in OpenType table. I am wondering if we could add a note to suggest how this value could be computed from font ascent and font descent when it's missing.

ideographic baseline has the same issue.

[drott]

I agree this should be addressed. UA's implementing different heuristics when the hang entry is not available from the base OpenType table is not interoperable.

[drott]

CC @kojiishi

[drott]

Compare related CSS WG issue #5380 for a similar discussion on what to do when a metric is not available.