#microformats 2022-02-01

2022-02-01 UTC
jacky, Seirdy, ur5us, zachburau[d], angelo, gRegor, KartikPrabhu1, bg2, [aciccarello], jamietanna, Loqi and diegov joined the channel
#
@threedaymonk
↩️ it probably did: it helped Google et al index websites, then they got big enough to bludgeon everything with ML, at which point microformats became more of a threat to them
(twitter.com/_/status/1488470632782340097)
zack[m], [jackjamieson], marksuth[d], diegov, mambang[m], juanchipro[m] and jacky joined the channel
#
jacky
I don't think https://microformats.org/wiki/microformats2-parsing covers this but if I had an element with multiple e-contents, should I merge them together or just keep going with the "every property value is a list" and just append it?
#
[KevinMarks]
I think existing ones will make a list, but try them and see.
#
sknebel
the parser does not care. everything is a list
#
sknebel
what you do as consuming code is a good question
#
sknebel
but I dont htink as a publisher you can or should rely on anything if you do that
#
sknebel
I think my code usually just uses the first and ignores the rest
#
jacky
hmm okay
#
[KevinMarks]
Joining them usually makes sense. You can wrap them in markup if that fits (eg join with <p>)
#
sknebel
yeah, but I dont think we have a documented "standard" on how to handle that
jacky, Guest6 and KartikPrabhu joined the channel
#
[tantek]
jacky, microformats2-parsing does not specify anything special about parsing e-contents compared to any other e- property.
#
[tantek]
[KevinMarks] I'm not sure I'd agree with "usually makes sense" unless you’ve written e-content consuming code, have joined them, and seen enough examples of that to determine that they "usually" make sense.
#
[tantek]
I could see it *possibly* making sense, as in theoretically.
#
[tantek]
right again sknebel, and we wouldn't until someone demonstrated a full publishing & consuming code scenario for that that "makes sense"
#
[KevinMarks]
I was thinking of the things I've done in unmung in various places, and combining the multiple parts of an item that way works, and if you do it with an element with the class on then it will preserve the original segmentation. But fair point, it does need some experimenting. In general joining seems better preserving of markup intent than taking first or last with that class.
#
jacky
yeah I think for now, I'll keep it as a list but if I somehow get bitten by it a lot in the future, I'll merge it
#
[tantek]
jacky, it could be for example, used to publish multiple languages of the same content
#
[tantek]
in the same h-entry (since other data, published, author, etc. would be the same for someone who is multilingual)
#
[tantek]
joining multiple translations wouldn't make much sense
#
[tantek]
"paging" it might (like in a swipe interface)
#
[tantek]
note that this is how multiphoto posts work with multiple u-photo properties in a single h-entry
#
sknebel
oh, thats an interesting point, yes
#
sknebel
we dont really expose per-property language yet
#
sknebel
but in principle...
#
jacky
so that's _partly_ related to why I ask but I didn't think it was relevant
#
jacky
b/c my parser has a `lang` field for content but it defaults to nothing if not provided (and whatever `html lang` is set to)
#
jacky
but like y'all mentioned; if it's not being published often, no real need to worry about it
#
jacky
I kinda wish we had some sort of 'microformats publishing in review' of what properties were used the most, the least, what were some new ones, etc
#
jacky
we could probably pull that from the sites listed in indiemap
#
sknebel
which it seems actually now has two implementers already
#
Loqi
[voxpelli] #3 Parse language information
#
sknebel
but a third would be welcome for feedback
#
sknebel
(either way with two implementations we should look at moving that forward)
#
jacky
ah nice!
#
sknebel
(but its complex enough that more feedback would useful)
#
jacky
yeah once I can get this rust impl going, I'd def put it up there
#
sknebel
so if you implement something with language, try that? :)
#
jacky
totally ;)
#
sknebel
(thats actually what I was hinting at when I mentioned the "per-property" above, had misremembered the status)
ur5us and jacky joined the channel
#
[snarfed]
(that's only classes that parsers interpret as "type," though, ie only h-*)
#
[snarfed]
it also has all u-*, https://indiemap.org/docs.html#schema-links , but sadly not every single possible mf2 class. could definitely be extracted from the HTML corpus though!
gRegor joined the channel
#
jacky
perhaps yeah!