Re: text-decoration-skip-ink auto should continue past behavior - 30+ years of underline behavior changed by latest CSS draft

On 02/22/2018 08:48 PM, Plumb-Larrick, C. Andrew wrote:
> Okay. You don't like it. So set the property the way you want it on your sites or in a user stylesheet. In all your posts I 
> haven't really seen an argument on the merits about this -- just about your preference (hey, that's what css directives are 
> for!) And a sort of out-of-context caricature of the argument for bias against changing defaults.
> 
> My impression is that underlining is relatively rare in professional typography outside of the Web, where best-practices 
> generally confine its usage to hyperlinks. (Otherwise, it is usually a substitute for italic, originally driven by the 
> limitations of typewriters.) So I haven't been informed by the research into existing typographic practices that the committee 
> has done, and that inform their draft. But I have no reason to doubt it. (I think *if* you have the germ of a good argument it 
> would have something to do with this special case of link-marking calling for different behavior than normal typographic best 
> practices.)
> 
> Because of this relatively limited use case for underline, I'm also relatively unconcerned about the outcome. On this side, 
> noting only the need for an eye toward the concern (stated by others, not you) about some 'non-ink' underlines looking like 
> two links. I think this is mostly unlikely to present an issue (such breaks will usually be within one word, and there are 
> generally other link cues like hover colors, etc). But it IS a very useful point and highlights something for designers to 
> attend to in the real world.
> 
> In light of what has been stated to be the research from underline usage in typography, I have no reason to doubt that the 
> proposed default for this property is the correct choice on balance -- particularly in light of the vast overall improvement 
> in browser typography support that we've seen.

I think a related problem is also quality-of-implementation. If the browser skips
too far on either side of the descender, the resulting underline looks disjointed
and is harder to distinguish as an underline; the position of the underline affects
how badly the line is disrupted. Etc. There's a parallel thread about this on the
blink-dev mailing list right now.

The CSS specs currently leave these decisions up to the UA: it's allowed to skip,
but not required, and the details of the behavior are unspecified:
   https://drafts.csswg.org/css-text-decor-3/#line-decoration

~fantasai

Received on Wednesday, 28 February 2018 06:49:42 UTC