#2838 Windows: 6.0.3: The obfuscation of planes-1,2,3 Unicode characters is not needed anymore in Windows 11⸣s console

[Ilya Zakharevich]

Composing another bug report today, I found that the CLI of gnuplot with set encoding utf8 would ruin the plane-1 Unicode char in ☺😊 — when one tries to input these from keyboard (my layouts — and fonts — allow this. As well as “much more” — to put it mildly).

Instead, ☺�� is shown. The code to replace the UTF-16 surrogates by � can be safely disabled now (at least when the surrogates come in correct pairs).

Thanks,
Ilya

[Ilya Zakharevich]

Just in case — for people who never saw it before: the support of these chars¹⁾ is backwards compatible — they take 2 UTF-8 surrogates to encode, and take the “2 cells worth” of the screen real estate to be shown. So the readline code does not need to be changed.²⁾

¹⁾ … and of combining marks — but those “do not always work”; I could find neither rhyme or reason with combining marks refusing to be placed correctly time to time!

²⁾ Hmm, maybe I’m too optimistic: one needs to be careful to not split those for partial display of the too-long lines! But gnuplot⸣s readline does not seem to support this intelligently anyway…