#microformats 2014-07-09

2014-07-09 UTC
KevinMarks joined the channel
#
tantek
KartikPrabhu: good to know
emmak, robmorrissey, tantek, KevinMarks2, paintedbicycle and gRegor` joined the channel
#
tantek
as in I hCard therefore I am?
#
tantek
wait - did autocorrect type that "hCard" then?!?
robmorrissey, eschnou, KevinMarks2, Soopaman, chiui, Garbee, pfefferle, KartikPrabhu, Atamido and pfefferle_ joined the channel
#
tommorris
tantek: I’m not sure there are any use cases for using link elements as u- properties. but we should probably have a formalised discussion on the wiki about it.
#
Loqi
tommorris: tantek left you a message 10 hours, 9 minutes ago: what do you think of changing http://microformats.org/wiki/microformats2-parsing##a.u-x[href]%20or%20area.u-x[href] to "a.u-x[href] or area.u-x[href] or link.u-x[href]" ?
#
tantek
tommorris - the use-case so far is http://adactio.com/about/myself/ which only contains a logo for adactio in a link element, which could work with u-photo if we put class=h-card on <html>
#
tommorris
but in the case of parsing, if it makes sense within the semantics of HTML, I’m not too bothered about there not being strong use cases, so long as one could plausibly use it that way and not break things.
#
tantek
"makes sense within the semantics of HTML" is necessary but not sufficient
#
tantek
as we deliberately avoid encouraging invisible metadata
#
tantek
so for example, microformats2 parsing ignores <meta> elements
#
tommorris
so, allow invisible metadata iff compelling use cases can be presented?
#
tantek
well, allow *semi*visible data
#
tantek
iff etc.
#
tantek
the link icon is semi-visible because it shows up in browser UI sometimes, and when a user chooses "Add to Home Page" on a mobile device
#
tantek
as opposed to <meta> tags in general which are invisible always or nearly always
#
tommorris
my issue is I can’t imagine a scenario where someone would write, say, an hCard that intermixes link elements amongst other properties
#
tantek
tommorris <html class="h-card">
#
tommorris
HTML LS says that you can use the link element in three context: “where metadata content is expected” (which is?), in a noscript element that is the child of a head element, or something involving microdata/itemprop
#
tommorris
tantek: that makes sense
#
tantek
someone writes an h-card that *re-uses* not intermixes link elements that already have the content or infomation
#
tommorris
so, the use of the <link> element to point to PGP public keys that indieauth is encouraging
#
tommorris
<html class="h-card”><head><title class="p-name”>Tom Morris</title><link rel="u-x-gpg-public-key" href=“…" /> etc.
#
tantek
right
#
tommorris
writes it up for the wiki
#
tantek
tommorris - the other aspect that makes me want to fix this (as proposed in the !tell to you above) is that we are already parsing <link> elements for rel values
#
tantek
so it's odd to ignore them for properties inside h-* microformats
#
tommorris
I think mf2py is just parsing for *[@rel] rather than any particular elements with rels
#
tommorris
which is what the spec says.
#
tantek
really?
#
tommorris
edited /microformats2-parsing-issues (+932) "link elements and u- parsing question"
(view diff)
#
tantek
spec says "to parse a hyperlink element for rel microformats:"
#
tantek
not "all" elements
#
tommorris
well, the Talmudic scholar will respond: what counts as a hyperlink element?
#
tommorris
just <a> or <area> and <link>? ;-)
#
tantek
tommorris: you added a new issue to the resolved section
#
tommorris
heh, not enough caffeine.
#
tantek
serves the Talmudic scholar appropriately
#
tommorris
edited /microformats2-parsing-issues (+1) "moved from resolved to issues"
(view diff)
#
tommorris
and, yes, the Wikipedia administrator sucks at using wikis.
KevinMarks joined the channel
#
tantek
so basically there's the (weak) consistency argument (<link> is parsed for rel, so should also be parsed for u-* properties), and there's the *potential* (strong) real world use case argument (if adactio goes the <html class="h-card"> path and uses <link class=u-logo rel="shortcut icon …" href="…"/> )
#
tantek
edited /microformats2-parsing-issues (+485) "/* link elements and u- parsing */ note semi-visible rel=icon link, use code instead of kbd for markup"
(view diff)
#
tantek
!tell barnabywalters,KartikPrabhu update for your consideration: http://microformats.org/wiki/microformats2-parsing-issues#link_elements_and_u-_parsing
#
Loqi
Ok, I'll tell them that when I see them next
#
tantek
edited /microformats2-parsing-issues (+69) "/* link elements and u- parsing */ link to specific uf2 rel parsing"
(view diff)
adactio and RCheesley_ joined the channel
#
tommorris
edited /microformats2-parsing-issues (+143) "/* link elements and u- parsing */ comment"
(view diff)
ChiefRA, Soopaman, pfefferle_, adactio_, TallTed, ivc\zz, tobyink, MMN-o, KevinMarks, elux, globbot, danielfilho and edsu joined the channel
#
tommorris
KartikPrabhu et al. ^^
#
@legacypython
mf2py 0.2.1: Python Microformats2 parser
(twitter.com/_/status/486874797586907137)
#
tommorris
"legacy". :)
#
tommorris
Now worrying about all the other craziness that lurks inside the HTML spec.
paintedbicycle, jgarber, KevinMarks, gRegor`, tantek, Soopaman1, caseorganic, Rastus_Vernon and Exploter joined the channel
caseorganic, eschnou, jgarber, KevinMarks, robmorrissey, KevinMarks_, Garbee and KartikPrabhu joined the channel
#
Loqi
KartikPrabhu: tantek left you a message 9 hours, 48 minutes ago: update for your consideration: http://microformats.org/wiki/microformats2-parsing-issues#link_elements_and_u-_parsing
Soopaman, dwayhs, paintedbicycle, KevinMarks and robmorrissey joined the channel
Rastus_Vernon, KartikPrabhu and tantek joined the channel
#
kp
edited /microformats2-parsing-issues (+153) "/* link elements and u- parsing */"
(view diff)
#
tantek
KartikPrabhu: re: should the door be opened to hidden mf data? note that the door is *already* open to <link> via rel parsing, and the <link>s we're talking about are all semi-visible, not completely hidden
#
tantek
(some of that is in the issue text)
#
KartikPrabhu
yes... but nothing stops people from using it for any <link>
#
KartikPrabhu
best practice would be to restrict to semi-visible ones
#
KartikPrabhu
in any case my point being, that if there are enough use-cases then it can easily be added
#
tantek
KartikPrabhu: when you say " from using it for any <link> " - can you give specific examples that you're concerned about?
#
tantek
I'd rather not add any special casing - to keep the model simpler / more predictable (to authors and parser developers)
#
tantek
in fact, that desire for a simpler model is what's driving the "weak" argument of consistency with "rel" parsing
#
tantek
why should <link> *only* be parsed for "rel"?
#
KartikPrabhu
hmm fair enough
#
tantek
why should <link rel="in-reply-to"> work but not <link class="u-in-reply-to"> (presumably inside an <html class="h-entry"> on a post permalink page) ?
#
KartikPrabhu
it is true that if people want to put hidden data they can already do so
#
tantek
again - can you provide a specific real world <link> hidden data example that you're concerned about?
#
KartikPrabhu
tantek: it seems any example I can think of can use a rel instead of u-* so it won't be a valid example
#
KartikPrabhu
so now I agree that <link rel> and <link class=u-*> should be on the same footing
KevinMarks joined the channel
#
tantek
ok just waiting on barnabywalters' opinion then
#
tantek
KartikPrabhu: perhaps update your comment on the issue then
#
KartikPrabhu
aah yes. on it
#
kp
edited /microformats2-parsing-issues (+266) "/* link elements and u- parsing */ added in favour"
(view diff)
#
tantek
KevinMarks: would appreciate your feedback and analysis of this too: http://microformats.org/wiki/microformats2-parsing-issues#link_elements_and_u-_parsing
#
KevinMarks
hm. That is tricky as it is a tension between DRY and invisible metadata
#
tantek
KevinMarks: see detail about semi-visible rather than invisible
#
tantek
<link rel=icon> is not truly invisible - it's semi-visible
#
KevinMarks
historically we've eschewed link-rel form some things (tags) and not others. If you allow link-rel allowing u-* on a link makes sense, though there may be containment issues wiht parsers
#
KevinMarks
do you put the h-* on the <head>? or the <html>?
#
KartikPrabhu
KevinMarks: that just decides the scope of the h-*
#
tantek
KevinMarks: we don't disallow h-* on <head> or <html> right now - it's just that no one is doing it. (as in current parsers support it on both)
#
KartikPrabhu
tantek: i think barnabywalters uses h-* on html
#
tantek
I see <html class="h-*"> applying to currently published markup use-cases
#
tantek
I don't see <head class="h-*"> applying to any currently real world markup use-cases.