Kirat Rai

[eggrobin]

[171-C13] Consensus: Accept 58 Kirat Rai characters in a new Kirat Rai block that extends from U+16D40..U+16D7F for a future version of the standard, with properties as amended in discussion. (Reference L2/22-043R)

[176-C25] Consensus: Generalize conjoining behavior to include Kirat Rai vowel signs, and set the Grapheme_Cluster_Break property of the Kirat Rai vowel signs {E, AI, AA, O, AU} to "V". For Unicode 16.0. See L2/23-160 item 4.2.

[176-A83] Action Item for Robin Leroy, PAG: Set the Grapheme_Cluster_Break property of the Kirat Rai vowel signs {E, AI, AA, O, AU} to "V". For Unicode 16.0. See L2/23-160 item 4.2.

RMG tracking: https://github.com/unicode-org/utc-release-management/issues/28

[markusicu]

Preliminary script code: Krai

[eggrobin]

Checked consistency with Ken’s UnicodeData-16.0.0d6.txt and LineBreak-16.0.0d2.txt.

[eggrobin]

CI does not pass because of collation.

[eggrobin]

do you want to come up with them yourself and want me to review them?

I tried that (looking at the proposal and the assignments for other characters, with a couple of forays into the standard to figure out what was going on), but I am of course still mostly clueless about this, so please review carefully.

[markusicu]

Re the UCA Main failure:

• In the debugger, I see that it maps U+16D69 KIRAT RAI VOWEL SIGN O to collation elements corresponding to its Decomposition_Mapping 16D63 16D67, and because we don't have allkeys.txt data for them, the tool doesn't know what to do and barfs.
• I don't know why it lets other pull requests through despite lack of data.

I will look at Ken's past notes and see if I can propose some data.

[markusicu]

I added an initial sort order definition in unidata.txt here: 42bcaa8
@eggrobin please run your sifter and check in allkeys.txt.
@Ken-Whistler would you mind taking a look at my unidata.txt changes?

[roozbehp]

Partial review

[markusicu]

Looks like the allkeys.txt file generated fine 🎉

The "variable" punctuation characters are a little hard to make out because they are not in collation order, but the script block looks as expected (I think):

...
16ABE ; [.4F5B.0020.0002] # TANGSA LETTER ZA
16D40 ; [.4F5C.0020.0002] # KIRAT RAI SIGN ANUSVARA
16D41 ; [.4F5D.0020.0002] # KIRAT RAI SIGN TONPI
16D42 ; [.4F5E.0020.0002] # KIRAT RAI SIGN VISARGA
16D43 ; [.4F5F.0020.0002] # KIRAT RAI LETTER A
16D44 ; [.4F60.0020.0002] # KIRAT RAI LETTER KA
...
16D68 ; [.4F84.0020.0002] # KIRAT RAI VOWEL SIGN AI
16D67 16D67 ; [.4F84.0020.0002] # KIRAT RAI VOWEL SIGN AI
16D69 ; [.4F85.0020.0002] # KIRAT RAI VOWEL SIGN O
16D63 16D67 ; [.4F85.0020.0002] # KIRAT RAI VOWEL SIGN O
16D6A ; [.4F86.0020.0002] # KIRAT RAI VOWEL SIGN AU
16D69 16D67 ; [.4F86.0020.0002] # KIRAT RAI VOWEL SIGN AU
16D6B ; [.4F87.0020.0002] # KIRAT RAI SIGN VIRAMA
16D6C ; [.4F87.0020.0004] # KIRAT RAI SIGN SAAT
10000 ; [.4F88.0020.0002] # LINEAR B SYLLABLE B008 A
...

FYI @Ken-Whistler but definitely no urgency!

I see that there is not an explicit contraction for the recursive decomposition 16D6A --> 16D69 16D67 --> 16D63 16D67 16D67. Maybe need to tweak the sifter code to provide the full canonical closure here?

[markusicu]

I see that we now have a different CI failure:

Caused by: java.lang.UnsupportedOperationException: unknown reorderCode 168
    at org.unicode.text.UCA.ReorderCodes.getSampleCharacter (ReorderCodes.java:146)
    at org.unicode.text.UCA.ReorderCodes.getScriptStartString (ReorderCodes.java:154)

This depends on me working on the script metadata. I have most of that data available, but don't have time right now to work it in.

We definitely need to change the CI so that a UCA-tool failure does not block UCD progress.

[eggrobin]

We definitely need to change the CI so that a UCA-tool failure does not block UCD progress.

Indeed. On it.

[eggrobin]

Re our friends alpha dia ext: all the interesting signs are Alphabetic by their Gc so that is easy. It would probably make some sense to make the viramas Diacritic, but it is probably best to deal with that as part of https://github.com/unicode-org/properties/issues/195.

Done.

[roozbehp]

I investigated the issue of InPC thoroughly, and I now agree with Ken and Robin that we need InPC for the right side depdenent vowels of Kirat Rai. Will include them in my copy of InPC.txt and will figure out how to merge them into main.