The CLDR root collation is a minor tailoring of the UCA DUCET. CLDR collation/root.xml defines a tailoring that was intended to un-tailor the differences, to make a DUCET-compatible collation available.
This has not been maintained (the CLDR 22 "ducet" tailoring only sets attributes but does not re-reorder any characters), and it would take some work to create, test and maintain.
Some differences cannot be un-tailored (U+FFFE & U+FFFF special CEs), others are not so easy (extender characters vs. alternate=shifted vs. symbol/currency reordering groups).
It is not clear who if anyone uses the "ducet" tailoring. Also, there is no automatic way to get language-specific tailorings applied to the DUCET order rather than the CLDR root order. (One would have to concatenate the ducet+language rules and build a new collator.)
I propose that we remove the "ducet" order and update the docs (LDML main & collation parts) accordingly.
would become obsolete.
Milestone 23dres deleted