2013/HTML is your Data

From IndieWeb

HTML is your data was a brainstorming session during IndieWebCamp 2013.

Notes archived from https://etherpad.mozilla.org/indiewebcamp-htmldata β€” probably need cleaning --Waterpigs.co.uk 10:31, 24 June 2013 (PDT)


HTML is your data

2013-06-22 17:00-17:15

  1. htmldata
  2. indiewebcamp
  • Use cases for web data
  • What HTML can we use for them
  • What are the issues we run into?

Use-case: Feeds (what used to be RSS)

  • HTML: h-entry (individual posts, home pages as feeds)
    • Issue: last-modified. with RSS/Atom you can send if-modified-since and get back a 304 with no body, and not have to parse, or do any other tests

Use-case: authentication discovery

  • RelMeAuth / IndieAuth rel-me to an OAuth provider profile URL
    • Issue: providers dropping rel-me support (FB), or wrapping with their own URLs (Twitter tco)

Use-case: Read-it-later -> Pocket / Instapaper / Readability

Use-case: social-graph - identity clustering

  • rel-me symmetric link graph - supported by identengine.com (consumes)

Use-case: n-factor-auth

  • RelMeAuth with requiring multiple OAuth providers

Use-case: home page security (Aside: HTTPS vs HTTP)

  • https on your personal domain - WillNorris.com does this, is all https
    • SSL with SNI (extension to TLS which allows the client to specify which hostname they're trying to connect to, so the server can provide a certificate custom to that hostname). This is the only way we're going to have tons of secure personal sites on shared hosting (same IP address).
      • cheap web hosting is typically shared hosting (share IP address with strangers)
      • or you host multiple people on your domain, e.g. given.familyname.org
      • issue: SNI doesn't work on <= IE8 or IE9/XP. Need IE9+ not on XP.
      • issue: lack of support in client libraries, e.g. Haskell library only recently added support for SNI. Janrain hasn't updated their Haskell library yet. So can't log-in using any Janrain OpenID service.
  • support document for Persona has to be on https

Use-case: scientific paper citations

Use-case: streams of physical measurement data from new devices

  • trend: accept a REST API, and provide JSON
  • opportunity: this is how you should be providing a stream of data from devices
  • microformats2 library available for converting HTML with microformats to canonical JSON representation


Question:

  • How to return JSON version of your web page?
    • Example: add ".json" to any URL on aaronparecki.com
  • What about HTTP ACCEPT header? application/json - CONNEG
    • Do both?

Use-case: email address based discovery

  • from alice@example.com ->
  • WF: example.com/.well-known/webfinger/?resource=acct:alice@example.com
    • that returns JSON
      • what does this tell you?
      • rel=feed - URL
      • rel=openid.server - URL
    • but it could return HTML+microformats
  • could eliminate with:
    • just put the rel values on example.com
  • could eliminate the well known URI with a new rel value
    • rel="rels" href="rels.html"
      • rels.html has all the same rel/URL pairs as your WebFinger JSON

Use-case: backups from web applications

Use-case: exports from web sites/silos/application Examples:

  • Google Takeout - G+ export has HTML+microformats
  • Facebook export has HTML+microformats
  • Twitter archive has JSON that it renders clientside HTML+microformats

Use-case: import an export into your own site

  • i.e. transitioning from sharecropping to owningyourdata

Use-case: outlines

  • HTML: XOXO

Use-case: rating information

  • HTML: h-review

Use-case: self-tracking

Use-case: one endpoint for robot clients and human clients

  • ACCEPT: mime multipart
    • what would a browser do if you sent it a MIME multipart response?

General Issues

  • Parsing speed at scale of HTML data
    • use a proxy like pin13.net
  • "But what if I don't want to display the data on my web page?"

See Also