Post

atproto/@standard.site friends: any thoughts on this issue? This... <link rel="site.standard.publication" href="at://..."> ...is invalid HTML, per html.spec.whatwg.org/multipage/se..., because DIDs aren't hostnames. This causes problems with some tools. Issue here: github.com/sveltejs/kit...

one idea (though it's really only working around the 'DIDs aren't hostnames' issue) would be to use <meta> tags instead of <link>. that's how non-standard-but-widely-understood things like OG images are done, after all feels like the sort of thing to fix now, before problems become widespread

hmm, yeah I think the meta tag is a good fit. Just need to think about how to document the change. There's probably a small enough set of consumers that we can get it done pretty quick.

hear, hear! Next up: rkeys of type any ?

I think we're coming to consensus that it's a good idea yeah! The biggest criticism against imo is that the chance of collisions when you have multiple publications under a single repository is pretty real, and rkey's just avoid that entirely.

that drawback makes sense as someone working on tooling though, it's nice to see it coming along (woo!)

yes, TIDs are definitely a reasonable default. but if you know what you’re doing, and have a unique identifier already (e.g. a URL path that includes the publication base-path), then it’s so much nicer to just use that instead of having to derive a TID from that. In our library we now support both: