CI check failures...

CI check failures...

Ah, we are using an older JDK than the one I have on my machine…

Thank you so much for addressing my feedback.

Will you add mirroring to bidi-c in this PR as well, or in a separate one?

My guess is it'd be something like this:

[NeatNit posts code, redacted]

Edit: although you might want to add some indication that the character has been mirrored. Terrible mock-up to get the idea across:

But this would require looking up which code points are Bidi_Mirrored, which would increase code complexity. I'll let you make the appropriate considerations.

Will you add mirroring to bidi-c in this PR as well

I will deal with that separately, when I can find time.

Please don’t post code in issue comments, we can’t look at it anyway because you haven’t signed the CLA.

And the point of the bidi-c demo is not to be a layout engine (this is most obvious in that it shows unshaped Arabic and does nothing about rule L3), nor to illustrate which characters the font happens to support mirroring for, but to be a versioned demo of the algorithm as specified, so of course it needs to look up the Bidi_Mirrored for the correct version of Unicode and flag those characters to which rule L4 applies.

It is not clear that it would be a good idea to actually cause the characters to mirror (or rather to maybe mirror if the font supports it) by setting the dir attribute; the display really shows the characters in isolation, not the glyphs they would have (this is striking in Arabic where it is unshaped), so changing the glyphs can make it trickier to figure out who is who; in addition, for the characters that do not have a Bidi_Mirroring_Glyph, actual support for the RTL glyph is usually limited to specialized fonts, so the display would be unreliable.