[css-inline] Should initial-letter-wrap: all consider text-shadow or other effects?

[dbaron]

Based on the twitter conversation ending in this tweet (note that the first tweet in the thread cites a tweet that is itself a thread) I'm wondering whether initial-letter-wrap should consider text-shadow or other effects.

This example of a first-letter using a font that's somewhat like text-shadow:

(which comes from this tweet) seems relevant. This example can't actually be done with text-shadow, I think -- it would have to be done using a specialized font. But maybe there are similar examples that would use text-shadow?

Is it worth making implementations consider text-shadow for initial-letter-wrap, or is that too obscure?

[astearns]

This example can't be done without exclusions, as without exclusions there is no way to break up the four lines in the middle. I don't think it's generally a good idea to break lines like those four, but this example is particularly bad with these unhyphenated (and syllabically-challenged) breaks:
ex perience
expr essed

Aside from that, I can see the benefit to avoiding the ink from a text shadow. I'd probably want to see evidence that people were converting their type effects to images and using shape-outside to achieve this effect before asking browsers to bake it in.

[bradkemper]

I would expect the text-shadow to not take up space, just like box-shadow doesn’t. And use shape-outside:polygon() for when that isn’t enough (though that would be laborious if you had to do it for more that one paragraph).

[fantasai]

I agree with @bradkemper here. Shadows don't take up space in CSS, not even for expanding scrollable overflow. You can do this with shape-outside, or by using SVG to represent the L and calculating the shape via shape-image-threshold. Especially since this behvaior isn't particularly common, I wouldn't want to build it into the CSS model.