#Smithereen updates.
- Many desktop UI tweaks to get that nostalgic layout right. Especially in the friends section.
- I've rewritten most of how i18n works. The language files are now flat json objects and use "ICU message format" for things like plurals and such.
- Non-ASCII (IDN) domains now work in mentions and search. Additionally, the link parser has improved. It now checks bare domains against the list of existing TLDs before turning them into links.
@grishka @activitypub
This is very nice.
And very funny, cause in the past days, I have also rewritten i18n for redaktor.
Also JSON/CLDR/ICU, see https://dojo.io/learn/i18n/introduction
However, what I found most distracting was the german language and some other gendered language.
My main problem is that I can use ICU at runtime but not to format existing nls terms …
My main problem was gender and so I began to write a thing so that it declines nouns and adverbs.
The cases in german are pretty much. If you want to decline it Singular/Plural for 4 genders (incl. ambiguous), it would be 104 different cases (4x26 w. determined or und. or no article and the cases).
Just pushed what I did yet today, e.g. https://github.com/redaktor/widgets-preview/blob/master/src/_ld/kanbanNLP.js
English is much more easy and also nearly ready.
And then it adapts if e.g. the actor is a single female or an ambiguous plural or whatever.
Don't know how complicated e.g. Russian is …
@grishka @activitypub
This is very nice.
And very funny, cause in the past days, I have also rewritten i18n for redaktor.
Also JSON/CLDR/ICU, see https://dojo.io/learn/i18n/introduction
However, what I found most distracting was the german language and some other gendered language.
My main problem is that I can use ICU at runtime but not to format existing nls terms …