Three possible uses for IRIs

[colecmac at protonmail.com]

> > The most difficult part of what you outlined is the Unicode normalization,
> > which maybe not all languages have libraries for, and would also require
> > updating every so often. But it wouldn't be a requirement for clients at all,
> > just something nice to have.
>
> If a client has an unnormalized IRI, it needs to normalize it before sending
> it to the server.

Could you justify this? It's a good thing to have, but it feels like a big ask,
as Unicode support and especially things like normalization are not straightforward
in all languages. I don't see why it can't just be recommended and not required.

> > I assume you mean NFC normalization?
>
> Yes.  When I speak of normalization, I mean NFC normalization exclusively.

Sounds good!

> > What if the user named a domain/file/folder in a non-NFC way? Now does the server
> > need to support NFC as well, and apply it to vhost recognition or local file paths
> > to correctly match requests? That seems wrong. But so does the user entering
> > something visually identical to what the sysadmin typed, and things not
> > working.
>
> I'm okay with that just failing, as file names are not really part of text/gemini
> content.  The difference will be obvious to the admin by checking the requested
> URIs from the server log against the %-encoded names of the folders.

The issue is that admins are not the only ones who create folders and files.
Non-technical people will as well, and a bug like this will be very confusing.
Everything will look right, but it just won't work. However, I doubt this will
occur very often, and it's an acceptable tradeoff to supporting Unicode.


makeworld