[ufmace]

I'm pretty sympathetic to this. I want to like the idea of Gemini, but it currently doesn't have, and is difficult to impossible to build, anything to really draw users away from the conventional web.

I think the fundamental idea of a new protocol that's much more limited than the current HTTP + HTML + JS web platform still has merit, but Gemini is just too limited to get going. It's tough though to design something that's flexible and feature-full enough to be really useful and not too hard to implement, but not so much that it will inevitably be extended to be basically what the web is now.

Thinking about it a bit, I think the idea for required TLS and client certs as the only way to ID users is good, as is lack of support for arbitrary HTTP headers and JS. I think what really kills a lot of potential functionality is the total lack of support for sending any data back to the server by anything other than basically query params. I think it ought to be possible for somebody to build a Super-Gemini that just support some equivalent of POST and some simple equivalent of HTML forms in addition to what it had, and should still be pretty easy to implement.

[azornathogron]

See the "titan" protocol.

https://transjovian.org/titan/page/The%20Titan%20Specificati...

It's "intended" to post a whole page/resource, but obviously it could be used to send an arbitrary blob just like an HTTP POST request.

[ufmace]

Huh never mind before it actually works fine at home, must have been a pesky work network filter against it for some reason. I've never seen our work filters do that before though.

[ufmace]

That site appears to be completely down, can't even find much of anything under archive.org.

[translucyd]

That's actually pretty nice!