Data Collection Request: sample HTTP Link headers

[jgarber623]

If your website includes HTTP Link headers in its response headers to HEAD requests, could you run the following command (or similar depending on your OS), replacing $url with your website's address, and paste the results in a comment on this Issue?

curl --head --silent $url 2>&1 | grep -i Link: | pbcopy

[jgarber623]

https://sixtwothree.org:

Link: <https://assets.sixtwothree.org/>; rel="preconnect"
Link: <https://fonts.googleapis.com/>; rel="preconnect"
Link: <https://fonts.gstatic.com/>; rel="preconnect"
Link: <https://sixtwothree.org/webmentions>; rel="webmention"

[cweiske]

$ curl --head --silent cweiske.de 2>&1 | grep Link:
Link: <http://cweiske.de/stapibas/xmlrpc.php>; rel="webmention", <http://phubb.cweiske.de/hub.php>; rel="hub"

[aaronpk]

my website and my curl client speak HTTP2 so the header name was lowercase:

$ curl --head --silent https://aaronparecki.com 2>&1 | grep -i Link:
link: <https://switchboard.p3k.io/>; rel="hub"
link: <https://aaronparecki.com/auth>; rel="authorization_endpoint"
link: <https://aaronparecki.com/micropub>; rel="micropub"
link: <https://aperture.p3k.io/microsub/1>; rel="microsub"
link: <https://aaronparecki.com/auth/token>; rel="token_endpoint"
link: <https://aaronparecki.com/>; rel="self"

[aaronpk]

see also, the tests in: https://github.com/indieweb/link-rel-parser-php/blob/master/tests/BasicTest.php

[cweiske]

also see the parseLinks-Tests in https://github.com/pear/HTTP2/tree/master/tests

[Zegnat]

$ curl --head --silent https://sink.zegnat.net/ 2>&1 | grep -i Link:
link: <https://sink.zegnat.net/micropub>; rel="micropub"
link: <https://sink.zegnat.net/micropub>; rel="authorization_endpoint"
link: <https://sink.zegnat.net/micropub>; rel="token_endpoint"

I used to have this all as a single link header, but had to split it up because consumer support was limited… In the early days the command would have returned:

link: <https://sink.zegnat.net/micropub>; rel="micropub authorization_endpoint token_endpoint"

[jordangov]

Why only for HEAD requests? What if an API doesn't do very well and only implements the Link header on other methods?

Anyway... the CanvasLMS API that I'm dealing with right now uses them for paging:

Link: <https://[DOMAIN]/api/v1/courses/1/assignments?page=1&per_page=10>; rel="current",<https://[DOMAIN]/api/v1/courses/1/assignments?page=2&per_page=10>; rel="next",<https://[DOMAIN]/api/v1/courses/1/assignments?page=1&per_page=10>; rel="first",<https://[DOMAIN]/api/v1/courses/1/assignments?page=6&per_page=10>; rel="last"

[jgarber623]

Why only for HEAD requests? What if an API doesn't do very well and only implements the Link header on other methods?

@jordangov This gem is specifically for parsing HTTP Link response headers (and not for parsing <link> elements in the response body). 😀

[jgarber623]

@aaronpk @cweiske Thanks for the links to those test suites!

[jgarber623]

@jordangov Oh, now I see the rather fine point you've made. Updated the Issue's description!

[snarfed]

try https://indiemap.org/ ! gives you 2300 sites all in one place, easily query-able, and self serve. 😁

eg here's an example graph from https://indiemap.org/docs.html#stats :

instructions on how to write your own queries here: https://indiemap.org/docs.html#data-mining