Relative links vs scheme-less links

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>