724540 - investigate use of ICU charset conversion to replace code/data in intl/uconv

Status: RESOLVED WONTFIX

(Core :: Internationalization, defect)

Comment 1

[Jungshik Shin]

Inevitably, ICU's character conversion data is not exactly the same as Mozilla's. I'm pretty familiar with the differences between two (having worked on Mozilla's in the past) and can help if necessary.

Comment 2

[Henri Sivonen (:hsivonen) (not reading bugmail until 2026-07-27)]

We should be careful not to import any Encoding Standard violations from ICU. It's not at all clear that it's a good idea to start using any of ICU's conversion code.

Comment 3

[Anne (:annevk)]

If we do this we reduce the set of code bases that implement encodings for browsers to two. And the open source set to one.

Comment 4

[Henri Sivonen (:hsivonen) (not reading bugmail until 2026-07-27)]

I have prepared a queue of patches to carefully remove the parts of uconv that we don't actually use. It's pretty sad that at the same time we build ICU converters, including ones that are totally Web-irrelevant and that we shouldn't be building even if we migrated to ICU converters (bug 944348).

I would still rather remove *all* ICU converters from the build and make ours Encoding Standard-compliant instead of fighting the ICU upstream regarding Encoding Standard compliance.

Comment 5

[Simon Montagu :smontagu]

This particular bug is a definite WONTFIX. (Bug 944348 is indeed sad, but this is not the place)