String.localeCompare with sensitivity:"accent" works incorrectly for Turkish dotless/dotted "i"

[andrew-z]

• Version: 12.18.0
• Platform: Windows (64-bit), version 2004

What steps will reproduce the bug?

"Istanbul".localeCompare("istanbul", "tr-TR", { usage: "search", sensitivity: "accent" })

How often does it reproduce? Is there a required condition?

Reproduces every time

What is the expected behavior?

"I" is not a capitalization of "i" in Turkish, so it should return a non zero value.

What do you see instead?

This returns 0.

Additional information

In new (Chromium-based) Edge browser console, I see "-1" output instead.

On the other hand, "istanbul".toLocaleUpperCase("tr-TR") works correctly in Node and returns "İSTANBUL"

[addaleax]

You’ll need the full internationalization data package for that.

For Node 13+, it is already included in the default node binary, for lower versions, you could use e.g. https://www.npmjs.com/package/full-icu.

[andrew-z]

You’ll need the full internationalization data package for that.

For Node 13+, it is already included in the default node binary, for lower versions, you could use e.g. https://www.npmjs.com/package/full-icu.

@addaleax thanks! Loading full ICU fixed the issue.