There are new differences between the DUCET and the CLDR root collation, in particular ten new Tibetan contractions in CLDR that are missing from the DUCET. We should adjust the CLDR root "ducet" tailoring accordingly.
We might have to suppress contractions for some code points (0FB2 & 0FB3) and then re-add the ones that are in the DUCET.
Note: Full matching of DUCET behavior will also depend on table builder code not automatically adding prefix contractions, but that is out of scope for CLDR.
Apparently CLDR "ducet" has already been out of sync with the actual DUCET. Richard Wordingham writes:
For example, in DUCET (both 6.1.0 and 6.2.0d9), we have, when the
alternate setting is non-ignorable, U+10A7E < U+10A7F < U+10917 with
nothing between successive items. In the CLDR root, its tightest
bounds, for UCA 6.1.0 and 6.2.0 on 24 June at least, are U+111C7 <
U+10A7F < U+0060.
There are other changes [...]
Note: U+FFFE & U+FFFF are not supposed to be tailorable, so the "ducet" tailoring cannot revert them to the regular DUCET implicit CEs.
Replying to (Comment 1 markus):
> There are other changes [...]
In , it is noted that other changes include the ordering of:
U+1D371 and U+02D0 (tidying up of numbers and punctuation)
U+A782 and U+20A8 (moving of currency symbols to a single block)
U+FDF6 and U+FDFC (moving of currency symbols to a single block)
I've added the reason for the change by way of explanation to the raiser of .
Postponing. The "ducet" tailoring is currently empty; it will need a bit of work to establish in the first place, not just to add new differences.
Obsolete: The "ducet" tailoring was removed with .