#microformats 2018-11-04
2018-11-04 UTC
[eddie], [kevinmarks] and [tantek] joined the channel
[kevinmarks] joined the channel
# 
[kevinmarks] Property?
# 
aaronpk also see the tests I added there https://github.com/microformats/php-mf2/pull/204/commits/fe339c78fc6ca342f9ee93dad334307fa84c1d08
[jgmac1106] joined the channel
# [kevinmarks] That seems like a breaking change if people publish that way. Did you find new evidence that u- classes were causing actual problems?
# [kevinmarks] I saw the sonniesedge.co.uk post, but not the notes
[tantek] joined the channel
# [kevinmarks] Previous discussion: http://www.kevinmarks.com/u-means-style.html
# [kevinmarks] Ah, yeah. That is a problem. Though if you are using BEM or something it should be obviously different. The real answer is to add integration tests that parse the microformats, like mastodon did.
# [kevinmarks] Or if you are using a preprocessor build model, compile the property into classes. When you run the rest of the build pipeline
[pfefferle], [dominik], [tantek], [jgmac1106], [kevinmarks], [eddie], [grantcodes], [frank], jgmac1106, [cleverdevil], jmac, [sebsel], TheTemp, [Khurt] and [dave] joined the channel
# 
[dave] Sorry to disrupt, but if anyone is interested, that Chrome extension I worked on has merged in my updates: https://beatonma.org/app/microformats-reader
# 
gRegorLove Interesting discussion on property vs class
# gRegorLove The HTML 5.2 section linked there seems to be about the ARIA properties, though, which are prefixed with "aria-" e.g. `aria-labeledBy="id"`
# gRegorLove Added to notes
# gRegorLove I have mixed feelings about it. We've referred to mf2 as implementing lessons from RDFa and superseding it. http://microformats.org/wiki/rdfa
# 
KartikPrabhu not sure why class is a problem
# KartikPrabhu the "style" vs other stuff already exists in the class attribute anyway
# KartikPrabhu see the first note https://html.spec.whatwg.org/multipage/dom.html#classes
# KartikPrabhu microformats fit nicely into "encouraged to use values that describe the nature of the content, rather than values that describe the desired presentation of the content."
# gRegorLove Re-using RDFa properties seems to muddy whether it's an mf2 document or an RDFa document. At a certain point, what's our answer to "why not just use RDFa?"
# gRegorLove Yeah, I'd like some clarity on what the proposed authoring change is too. Is it "use property first, or class as a fallback" or something different.
# KartikPrabhu before the authoring change, is there any reason to not use class
# gRegorLove Summary of the discussion is the two reasons are: clashing with CSS naming conventions and devs removing HTML class names because they appear to be unused.
# KartikPrabhu those are not mf2 specific anyway
# gRegorLove But agreed, I'm not certain those are strong enough reasons. Look forward to reading more reasoning on it.
# KartikPrabhu devs could remove properties too
# gRegorLove Yeah
[stefp] and [jgmac1106] joined the channel
# aaronpk The argument is specifically that class is primarily used and understood as styling, so someone who sees a class name they aren't styling may remove it. In contrast, someone who sees a "property" attribute and doesn't know what it's for may actually think twice before removing it because it was clearly there for some reason rather than just being an unused styling class
# KartikPrabhu that is the complete opposite of what the HTML spec says for class attribute
# KartikPrabhu you can use any attribute for styling
[schmarty] joined the channel
# KartikPrabhu well then, the "best" way to be fool-proof is to invent a new attribute like microformats = "h-entry"
# KartikPrabhu and there we have it. we are into RDF land
# KartikPrabhu <sigh>
# KartikPrabhu if specs are not very relevant then neither are validators
# KartikPrabhu in practice, you can put any attribute on any HTML tag
# KartikPrabhu gives up
# gRegorLove That's what I gathered from the notes. Currently I lean towards educating about the use of `class` vs introducing `property` in the mix.
# gRegorLove Unless it's really becoming a big problem. Maybe it is.
# gRegorLove Is it just the u- prefix or others?
# gRegorLove bootstrap4 has some too, iirc
# gRegorLove Gotcha
[calumryan] joined the channel
# KartikPrabhu not sure why these are conflicts. a mf2 parser can just parse p-px
# KartikPrabhu mf2 consumers won't be using it
# gRegorLove Ok, but which is a better solution? Direct devs not to remove things they don't understand, or change how to publish mf2? I'm not sure.
# gRegorLove The latter feels like a big hammer
# gRegorLove Isn't that up to a project team that decides to use mf2 to communicate?
# KartikPrabhu so this will be mf3 ?
# gRegorLove It just sounds hypothetical to me. If using mf2 is important they should communicate that to people touching the code.
# gRegorLove "should" is hypothetical too, I realize, heh
[jgmac1106] joined the channel
# 
[jgmac1106] [aaronpk] wonder if she would do a microcast with/without you or post on it.
# [jgmac1106] Probably would solve 90% of WordPress problems... A totally arbitrary number based on an over estimation of my knowledge
# [jgmac1106] Also I can't remember. Did I get the link to the NYC building block session to you?
# 
Zegnat I think Charlie’s point was mostly: here we have a somewhat bigger scale institute than just private sites, and people with an interest in marking things up semantically. But the people with that interest are not the developer team, or possible outside front-end workers. At which point you have very few choices: educate the entire staff, or not do it at all.
# Zegnat So she came into the session basically asking what she could use that wasn’t microformats, as the signal was clear that adding microformats the way currently specified was just never going to happen. So Microdata, RDFa, and JSON-LD, all came to mind, but we also wanted to look into alternative ways to express the mf2. (One thing also mentioned: the idea to give WP pages separate URLs that already provide mf2-json.)
# Zegnat This led to the question, what if we just do not use the class attribute? What other attributes are there? We did a lot of checking in the spec and there is no clear custom-attribute way supported by HTML5 (other than data-* prefixed ones, who have a different defined purpose). Then someone (I think tantek) mentioned that OpenGraph is using a `property` attribute. And at the time in the room nobody really seemed to have a reason why
[kevinmarks] joined the channel
# gregorlove edited /history-of-microformats (+139) "/* 2010 */ +microformats2 proposed and conceptually explored during a discussion session at FOO East 2010." (view diff)
# [kevinmarks] I would like to have heard Charlie talk about that, and what the tool chain was that makes this likely. The reason we ended up with prefixes initially was to avoid the arbitrary class problem.
mblaney joined the channel