Post #3824692

2026-07-15 01:51 UTC

a nice unexpected benefit of godbolt.org is that it's making me aware of more system languages. today, i learned about the C3 language via the godbolt language pull-down menu. for prototyping purposes (not for production, which requires memory safety), C3 seemed to check a lot of boxes. but then i got to the part where it milkshake-ducks itself: https://c3-lang.org/faq/rejected-ideas/#unicode-identifiers they might as well have asked, "why would anyone think in their native language, even in their personal experimental code?"

Replies (4)

@cait@anarres.family 2026-07-15 04:44

@JamesWidman@mastodon.social the bit "there's no evidence anyone does this" is telling - there's never any evidence if you shut your eyes I sometimes use Welsh for variable names - I'm not even fluent, but it's a good way to improve my fluency (Ironically Welsh is poorly-served by Unicode: the language has its own distinct, well-defined alphabet with unique letters, but those letters are written as digraphs - dd, ll, ng are some examples - so they've never, to my knowledge, been admitted to Unicode as code points)

Open ##3824691
@JamesWidman@mastodon.social 2026-07-15 02:01

unicode mavens, please correct me if i'm wrong, but this: > "Furthermore there are practical issues, such as bidirectional text [...]" does _not_ seem applicable to the question of whether a _compiler_ should allow non-ascii characters in identifiers. (iiuc, you only need to worry about bidi stuff when you're rendering glyphs. and compilers generally don't have render pipelines.) so this seems like a bullshit excuse.

Open ##4317111
@pinskia@hachyderm.io 2026-07-15 02:41

@JamesWidman@mastodon.social Japanese code is written with huge amount of Japanese comments. All of the cell specifications were translated into Japanese for both Sony and Toshiba even if the original was written in English. Chinese code is all native in Chinese. I had to wait for someone to translate comments for me while I was trying to help aliaba improve their code. Including their open source code. The English omly crowd is hugely Americans and British folks who are have not learned any other language or hate on non native speakers. Even the push for llms for translations to English is another English only thing. Using a human translator is good, learning English is neutral. But pushing llms for translation is just bad.

Open ##4317115
@pinskia@hachyderm.io 2026-07-15 02:43

@JamesWidman@mastodon.social Oh p4 is another language which does not take into account non ascii identifiers into account. It was never even brought up as a problem. The main reason why many of these languages do that is because they depend on flex rather than hand written lexers.

Open ##4317116