As a super skeptical grouchy old man, something feels right about it. Not just from a technical perspective, but also the chances of it actually getting some traction and being a viable alternative to the js-laden corporate web.
Something I love about Gemini is how easily it maps to terminal usage, as opposed to something like using lynx for the Web, for example.
I created a terminal Gemini browser that does all the things I want, Amfora. You can also check out the software list, there a few other terminal browsers.
I can't get on board with Gemini because it insists that connections cannot be reused and they must be TLS. If it was just one of those then I would be more enthusiastic. But there is a lot of stuff that would be fine without TLS and would be much faster, and it's always going to be faster if you can reuse connections. And if people are going to ignore that part then it shouldn't be in the spec.
Personally what I would be more excited about would be a fast pub sub content-oriented distribution network with things like linked markdown documents and web assembly applications on top of it.
Or maybe a browser extension and distributed database that we could use to identify and certify web pages that load extremely quickly.
I really love it; it's so nice to know that every gemini page is a single text document I can read top-to-bottom, and navigation is greatly simplified by giving all links their own line. This gives clients room to display links in different ways: some clients show the full link and label, others show just the label, and some truncate links to their domains visible below labels. Just to give you a taste, I formatted this comment the way that some gemini clients display gemtext (sans color, which more clearly separates links from labels). This touches one of the core ideas behind Gemini: presentation is up to the user agent, not the author. This might have been true at one point on the Web, but it is far more prevalent on the Gemini space.
Browsers like bombadillo allow seamless mixing of the finger, gopher, gemini, and http(s) protocols to offer a protocol-agnostic single browser to consume the "small plaintext-consumable Internet"
And because this inevitably comes up on HN: defining a "sane subset" of the web misses this point because I don't know whether any given link will lead me out of the sane subset, and because said "sane subset" will likely end up re-inventing gemini in the first place.
I suggest first taking at the official project website and the project's (more detailed) gemini capsule for more information.
Servers MUST use TLS version 1.2 or higher and SHOULD use TLS version 1.3 or higher. TLS 1.2 is reluctantly permitted for now to avoid drastically reducing the range of available implementation libraries. Hopefully TLS 1.3 or higher can be specced in the near future. Clients who wish to be "ahead of the curve MAY refuse to connect to servers using TLS version 1.2 or lower.
And clients that want to test against the Gemini Client Torture Test [1] MUST use TLS 1.2 as the server there only supports up to TLS 1.2 (due to said library limitation).
