I was browsing in Bombadillo and just noticed something: If I follow a link without a scheme, it just assumes it's a relative link. This works fine when it actually is a relative link, but what about this? => makeworld.gq/gemlog/ Link to gemlog As humans, most of the time we would assume that to be a scheme-less link, but Bombadillo will actually try and resolve that as a relative link. And why not? It very well could be. I am just writing these wrong, and scheme-less links should always start with `//`? The spec says "a scheme of gemini:// is implied", so I assumed having those would be redundant. makeworld
> On Jun 2, 2020, at 20:49, colecmac at protonmail.com wrote: > > I am just writing these wrong, and scheme-less links should always start > with `//`? The spec says "a scheme of gemini:// is implied", so I assumed > having those would be redundant. Sometime, the simplest things turn out not to be so simple :) See RFC-3986, section 5.2.2, Relative Resolution: https://tools.ietf.org/html/rfc3986#section-5.2 And of course, all of Sean's torture tests: https://portal.mozz.us/gemini/gemini.conman.org/test/torture/
colecmac at protonmail.com writes: > link. This works fine when it actually is a relative link, but what > about this? > => makeworld.gq/gemlog/ Link to gemlog > I am just writing these wrong, and scheme-less links should always start > with `//`? The spec says "a scheme of gemini:// is implied", so I assumed > having those would be redundant. The // are part of the authority section of the URI, which includes the host. If this weren't the case, the intepretation of the URI would be ambiguous. To specify the host but not the scheme you need => //makeworld.gq/gemlog/ Sean's client torture test is actually a great source of information on things like this, and any client that passes it is going to behave in the same way. It'd be nice to have it made explict somewhere else too. plugd -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 487 bytes Desc: not available URL: <https://lists.orbitalfox.eu/archives/gemini/attachments/20200602/d7f5 be9f/attachment.sig>
Okay, I stand corrected. Thanks! makeworld ??????? Original Message ??????? On Tuesday, June 2, 2020 3:24 PM, plugd <plugd at thelambdalab.xyz> wrote: > colecmac at protonmail.com writes: > > > link. This works fine when it actually is a relative link, but what > > about this? > > => makeworld.gq/gemlog/ Link to gemlog > > > I am just writing these wrong, and scheme-less links should always start > > with `//`? The spec says "a scheme of gemini:// is implied", so I assumed > > having those would be redundant. > > The // are part of the authority section of the URI, which includes the > host. If this weren't the case, the intepretation of the URI would be ambiguous. > > To specify the host but not the scheme you need > => //makeworld.gq/gemlog/ > > Sean's client torture test is actually a great source of information on > things like this, and any client that passes it is going to behave in > the same way. It'd be nice to have it made explict somewhere else too. > > plugd
By the way, in the WHATWG URL spec, the standard name for a ?scheme-less URL? is a ?scheme-relative-URL string?: https://url.spec.whatwg.org/#scheme-relative-url-string -------------- next part -------------- An HTML attachment was scrubbed... URL: <https://lists.orbitalfox.eu/archives/gemini/attachments/20200603/83f1 7ce5/attachment.htm>
---