lusty void
This is my first Gleam library 😎: https://github.com/richard-viney/iso_8859
It decodes ISO/IEC 8859 text content to a native Gleam string. When targeting JavaScript it falls back to the built-in TextDecoder which is faster.
It's fairly small/niche. I've got a much larger library in the works and decided to extract this small part of it as a trial run for actually publishing something!
Any feedback on it would be gratefully received.
steady yew
TIL you can have both an implementation and @external for a function
why is there a @external(javascript, "./text_decoder.mjs", "bits_to_codepoints") that leads to a function that just throws?
a thought: if the LUTs are not used in the JS target, they could be moved off to their own module (perhaps internal if you don't want them to be shared) so that a browser doesn't need to load that data. or could constants be marked with @target or something? I forget
that LUT part is genious btw 🙂
lusty void
Managed to remove that empty external by not specifying unsigned in the bit array pattern match (not supported in JavaScript currently but it's the default so can do without in this case)
Might be possible to not have the LUTs in JS, true, not sure if it's possible tbh
steady yew
ah, I realise the browser would import the constants anyway, so just moving them to another module isn't a solution
lusty void
These single-byte character encodings are all pretty simple mappings, so some bit array LUTs made sense, should be faster than a List(String) but I've not benchmarked it
steady yew
as a Finn I have plenty of personal experiences with ISO-8859-1 and -15 pain 😄
lusty void
Oh is it not largely UTF-8 these days?
In my case I'm dealing with binary formats that have been around for decades and support various non-UTF-8 string encodings
steady yew
it is, these days. 😛
glossy citrus
I've got a much larger library in the works and decided to extract this small part of it as a trial run for actually publishing something!
multiple small libraries is quite cool
steady yew
I remember using this sort of thing on IRC where half the clients were speaking ISO-8859-15 and half were speaking UTF-8 🙂
magic flower
I use a TextDecoder for keyboard input in teashop.
So this might be helpful for non-JS support for me.
fresh dirge