should parsers support include-pattern and friends

[willnorris]

microformats v1 includes three methods for including properties and child microformats by reference:

• include pattern, with class="include" and href="#ref"
• microdata itemref
• td "header" attribute

Test cases exist for all three of these at https://github.com/microformats/tests/tree/master/tests/microformats-v1/includes.

The mf2 parsing spec makes no mention of supporting include-pattern in its backwards compatible, however the existence of test cases indicates to some implementers (namely, myself) that maybe it is expected to be implemented?

Brief discussion on IRC (starting here) revealed that several folks were unable to find an example of include-pattern being used in the wild.

So the questions are:

• should any of the three inclusion methods above be supported by mf2 parsers? Either way, should that be explicitly called out in the parsing spec?
• if not, should the tests be removed? (Since there are likely no mf1 parsers in active development, I think it's probably safe to limit the test suite to only include mf2 parsing tests?)

[Zegnat]

Of the 6 example uses in the wild, only two seem to still be online and none of the microformats2 pages links to the include pattern.

My personal take is that, unless we are able to dig up more use cases, dropping the tests makes sense to me. Especially now that the tests are getting cleaned up and parser developers can start actively basing their work on it again.

We could add a special heading block to the include pattern wiki page clarifying that it isn’t currently part of the microformats2 specification and link to this (or a more specific issue) where we coordinate discovery.

I sometimes hear that my issues and concerns with the spec are too theoretical and not based on actual usage. So this time I’d like to argue the exact opposite :wink: I don’t see a reason to start shoehorning this into the parsing specification just because it used to be a thing years ago. Especially seeing how several parser developers have been able to tinker on parsers for years without realising this was even a thing.