aaronpkfor example most people don't include the links to syndicated copies of posts in the list view, but they do on permalinks (tantek is an example off the top of my head)
Zegnataaronpk, snarfed: I recall having seen discussions like this several times. Most recently as part of salmentions. I think some people were also talking about always fetching the full comment origin page and parsing that instead of relying on the comment display on a third-party website.
ZegnatThe general consensus always seemed to be that the only place to parse a mf2 object on is on its permalink. But that obviously makes any reader implementation more involved as just parsing an h-feed might not do.
ZegnatThere is also the problem when people don’t add a u-url to their h-feed. No way to actually know if an h-feed you find in the current HTML is the main object at that page’s URL. What if there are multiple h-feeds, like on tantek’s page?
ancardaHi everyone, I'm working on a WebMention implementation for my site. It seems like it's a good idea to save all mentions that have been received into a database. Are there any good limits on the length of source, target, and the overall POST body? I'm unsure how big to make the VARCHAR columns
ZegnatPeople have played with Mastodon in general, I think. And there is http://fed.brid.gy/ to try and communicate between IndieWeb websites and social stuff like Mastodon
aaronpkI decided to skip the step of /authorship that involves making another GET request entirely, since that would potentially mean I'd be making a GET request for each item in the feed
aaronpkso if there's a top-level h-card like tantek, then that h-card becomes the author for each entry. but it won't go and fetch data from another URL if it can't find anything otherwise
aaronpk"Wired magazine published an article called "The Curse of Xanadu", calling Project Xanadu "the longest-running vaporware story in the history of the computer industry".[2] The first attempt at implementation began in 1960, but it was not until 1998 that an incomplete implementation was released. A version described as "a working deliverable", OpenXanadu, was made available in 2014."
ZegnatWait, where in that original ideas list on Wikipedia does it mention that pages need to tell other pages when they link to them? Because that’s what Webmentions are :/
[kevinmarks]Previous version of that discussion. Sample quote Clay Shirky: "If you thought that all that was wrong with RSS 1.0 is that RDF didn't make it confusing enough, attach the twin boat-anchors of OWL and the Semantic Web to [Atom] and see what happens."
snarfedaaronpk: i was thinking about your partial feed content problem. you might consider just skipping it in v1, instead of always fetching permalinks. seems reasonable for a first pass MVP, esp if it's easy to click through. you can always add full fetching later.
ZegnatYeah, that is still the recommended mark-up for notes, I think. Personally I am not a fan of forcing every object to have a name. We might want to take that to #microformats though
ZegnatWe debated that at IWC Berlin: using fragment URLs as permalinks for posts in a feed. I recommended against it as I didn’t know any parsers that would understand it. Guess XRay does :D