|
|
|
|
|
by rurban
117 days ago
|
|
That's a user pipeline problem. If you just check confusables without any tr39 algo you will be disappointed also. I had to patch confusables for the C/C++ proposal for about 10 characters also. "...the default confusables list is extremely buggy. It needs at least 7 manual exceptions for the ASCII range, 12 exceptions for Greek, and I didn’t check any others scripts. python and clang-tidy were very unsuccessful with this approach, compared to java, rust and cperl with the mixed-script approach." https://rurban.github.io/libu8ident/#confusables In detail:
https://rurban.github.io/libu8ident/doc/D2528R1.html at 10 TR39 Mixed Scripts |
|