I expect that most (if not all) of the Unicode utilities in intl/unicharutils can be replaced by calls to ICU - either directly, or via lightweight wrappers if we want to maintain existing APIs for the rest of Gecko or for use by extensions, etc.

This does mean that every browser except Microsoft ends up with the same underlying implementation and bugs :/

We have already switched a few of the functions in nsUnicodeProperties over from our own data tables to ICU (in builds where ENABLE_INTL_API is defined). We can trim a further 80K or so from libxul by using ICU property accessors to replace most of the data in the nsCharProps2 array.

https://hg.mozilla.org/integration/mozilla-inbound/rev/4e53666d77dc54ead9ca93c38041ebf7cfdb6848 Bug 724538 - When ICU is available in the build, replace most of nsCharProps2 fields with ICU property accessors. r=emk https://hg.mozilla.org/integration/mozilla-inbound/rev/2038722d99d4d73a5ff508f20e51c7251c267a8d Bug 724538 - Regenerate Unicode property data with updated script. r=emk

https://hg.mozilla.org/mozilla-central/rev/4e53666d77dc https://hg.mozilla.org/mozilla-central/rev/2038722d99d4