Our `wcwidth` library is out of date

[tig]

We use https://github.com/spectreconsole/wcwidth under Rune.GetColumns which is a port of https://github.com/jquast/wcwidth.

This library is out of date and returns incorrect information for some codepoints.

In #4255 I've implemented a hack/workaround.

This issue is to track fixing this correctly, either by getting https://github.com/spectreconsole/wcwidth fixed, or some other means of getting a dotnet wcwidth library.

[tig]

Once the /spectreconsole/wcwidth nuget package is updated to 3.0 we can remove the temporary fix I put in which will close this issue.

See spectreconsole/wcwidth#10 (comment)

[patriksvensson]

I have released version 4 of wcwidth yesterday that adds support for zero width joiners (which adds support for emojis like 👩‍🚒), and VS16 (narrow to wide variation-16 selectors) among other things.

Wide character lookup logic has also been improved to cover more cases (and fix previous faulty ones).

[BDisp]

@patriksvensson thank you for your promptness in informing us of the new version. However, I see that there are some errors in the unit tests that lack a plausible explanation and that did not occur before.

The code point for the Devanagari character "कि" is U+0915 for the base consonant "क" and U+093F for the vowel sign "ि" (i-kar). When combined, these two Unicode characters form the ligature "कि".

• क (ka): U+0915
• ि (i-kar): U+093F
• कि (ki): U+0915 + U+093F

Should be expected a width of 2 but return 1.

Edit:
It's was returning a width of 2 because it was returning the sum 1+1=2 for each rune. Now after my PR #4383 it return now a width of 1. Maybe could be debatable but the Unicode table consider both runes as narrow.

Update:
It must really have a width of 2 because with the new PR #4388 the grapheme should return the sum of all code points until the maximum of 2 columns. If a code point doesn't occupy any cell when combined with another code point, then it must have a zero length.

Hangul Jungseong A (ᅡ): U+1161

• This is Other Letter category.
• This is a "middle" or "jungseong" character, which combines with a "consonant" or "choseong" character to form a syllable block.

It's a visible code point and should be expected a width of 1 but return 0.

Hangul Jungseong O-Yeo (ힰ): U+D7B0

• This is Other Letter category.

Should be expected a width of 1 but return 0.