On Wed, May 6, 2020 at 6:01 AM Corentin Jabot via SG16 <sg16@lists.isocpp.org> wrote:

We know that while arbitrary concatenation if NFC sequences may not be NFC, concatenating NFC identifiers is always NFC because XID_Start combining are never combining.

The entire Hangul Jamo block is XID_Start:

1100..1248 ; XID_Start # Lo [329] HANGUL CHOSEONG KIYEOK..ETHIOPIC SYLLABLE QWA

The table in the section (https://unicode.org/reports/tr15/#Concatenation) that you referenced has an example requiring only characters from that block:

NFC: {U+1100}

NFC: {U+1161}{U+11A8}

Not NFC: {U+1100}{U+1161}{U+11A8}

[ ... ] Notably there should not be isolated universal-character-name that are not well-formed identifiers.

These are technically required for the description of tokenization to be clear. That's all we use these for. Their presence immediately makes the program ill-formed (CWG folks would insist that I indicate "unless if UB exists anyway").