#microformats 2018-01-13

2018-01-13 UTC
KartikPrabhu, iwaim___ and [keithjgrant] joined the channel
#
[keithjgrant] hm, xray is interpreting my h-recipe as a review https://xray.p3k.io/parse?url=https%3A%2F%2Fsidecar.keithjgrant.com%2Fdrinks%2Ffrench-kiss%2F
#
[keithjgrant] looks like I have some weird <p> tag formatting inside it. that could be part of it.
#
aaronpk that is weird
#
aaronpk hah found the bug
tantek joined the channel
#
gRegorLove Closed two of those php-mf2 issues. Commented on the third; need some feedback/direction: https://github.com/indieweb/php-mf2/issues/60#issuecomment-357400495
#
Loqi [gRegorLove] @aaronpk Let's leave this open for now. I read https://github.com/indieweb/php-mf2/issues/60#issuecomment-228587873 as "authoring issue, ignore." However, I just tested the original HTML with #131 and it parses a space character for `updated`. I'...
#
gRegorLove tantek, Zegnat, sknebel ^
#
gRegorLove aaronpk ^^
#
gRegorLove Node parser supports YYYYMMDD format, despite spec
#
KartikPrabhu I propose a MYYDYMYD format
#
gRegorLove mf2py returns empty array
#
tantek more like "beyond spec"
#
gRegorLove I propose you shu-- jk, KartikPrabhu ;) :D
#
KartikPrabhu :P
#
gRegorLove Typo. Meant: you're super
#
gRegorLove tantek: aye, that's better phrasing
#
gRegorLove Ruby parses "2014-12-04 00:00:00", adding time precision not authored.
#
tantek gRegorLove: I'm not sure I understand the state of the issue
#
gRegorLove Hehe, and Go gets the string inside the span: "4 December 2014"
#
gRegorLove Just a moment
#
tantek re: supporting YYYYMMDD - that's likely harmless
#
tantek and if multiple parsers decide to adopt it, then that makes a case for adding it to the spec to reflect implementation reality
#
gRegorLove My main question is: if parser doesn't support YYYYMMDD, should the property be ignored, or parsed as an empty string?
#
KartikPrabhu good question
#
gRegorLove Agreed it seems like a harmless enhancement to support the format. php-mf2 doesn't currently
#
gRegorLove Ruby issue filed: https://github.com/indieweb/microformats-ruby/issues/76
#
Loqi [gRegorLove] #76 Don't add time precision unless authored
[miklb], tantek, chrisaldrich, gRegorLove, nitot, [kevinmarks] and [mrkrndvs] joined the channel
#
Zegnat Procrastination of the day, 2 issues files on value-class-pattern :)
#
gRegorLove If I could get a review and some thumbs up (and no objections) on https://github.com/microformats/microformats2-parsing/issues/17, I'll update the spec
#
Loqi [Zegnat] #17 Define removal of SCRIPT and STYLE elements everywhere textContent is requested.
#
Zegnat LGTM
#
gRegorLove I'm probably in favor of deprecating value-title parsing, but YYYY isn't a valid format according to VCP date and time parsing.
#
gRegorLove Which does seem a bit odd, but I don't know the history of VCP
[miklb] joined the channel
#
gRegorLove Haha "Why might someone write some markup like this? Ask a consulting psychologist." in the VCP questions section.
#
Loqi ahahahaha
#
Zegnat Oh, I guess I should change the dt- in the test to p- then and rerun the parsers.
#
Zegnat I just assumed this one was allowed without rechecking vcp date time parsing
#
Zegnat Honestly I was just rewriting vcp for myself so I would know how I would want to implement it, and these 2 issues came up. See https://wiki.zegnat.net/media/mf2dom.html
KartikPrabhu joined the channel
#
gRegorLove So close on this backcompat phpmf2 backcompat update, just one test giving me grief.
#
gRegorLove .h-entry > .hfeed > .e-content is a pain, heh
#
Zegnat Alright, why is that specific one a problem?
#
gRegorLove Hard to explain. Method I'm working on is recursive, digging down to each mf root. It parses them for properties correctly based on backcompat or mf2. Upgraded backcompat elements are kept track of.
#
gRegorLove .e-content gets ignored in the hfeed correctly, but gets caught by h-entry later, so the content property shows up.
#
gRegorLove I need to retrace my steps; this test was passing the other day. I changed something to break it :)
#
gRegorLove This is https://github.com/microformats/microformats2-parsing/issues/11#issuecomment-352281134
#
Loqi [sknebel] We just encountered this difference in handling between `mf2py` and `php-mf2` while trying to help debug Bridgy on a WordPress site with the following structure: ```html <body class="h-entry"> <div id="page" class="hfeed site wrap"> <h1 cla...
#
Zegnat Hmm, I’d have to look at your current parsing code to comment I think. Not in a very clear code mood tonight though.
#
gRegorLove No worries, I'll find it.
#
gRegorLove yay, I think I just did.
#
gRegorLove hard is recursion
#
Zegnat One of the reasons I tried to write down the DOM based algo for VCP today :P
#
Zegnat But I think you said you were doing XPaths?
#
gRegorLove Yeah, php-mf2 has always used xpaths. I haven't deviated too much from that
#
gRegorLove But now it's more precise: drill down to the most-nested mf, run backcompat to upgrade elements as needed, then use xpaths to extract the p-, e-, dt- etc. and parse them accordingly.
#
gRegorLove Before it was more like, find a root mf, if it's an mf1 root run backcompat on anything inside it.
#
Zegnat Interesting, so you go through the entire DOM at least twice? Once for upgrading backcompat, then for actual mf2 parsing?
#
Zegnat He, I just realised my DOM algo for VCP can’t be literally copied into PHP as PHP’s DOM doesn’t seem to have a TreeWalker.
#
gRegorLove Probably more than that, though it's not traversing every element. It's using xpaths a lot. If you have an hentry, it runs an xpath for each of hentry's properties to upgrade those elements.
[cleverdevil] and tantek joined the channel
#
Zegnat Hmm, interesting. I wonder if a pure DOM traversing lib can be faster. That’s a project for next month though.
#
gRegorLove It could probably be more efficient, for sure. No noticeable slowdown, though. Hoping this should fix 99.9% of weird mf2/mf1 combinations; definitely giving more correct results so far.
#
Zegnat Woo!
#
Zegnat gRegorLove++
#
Loqi gregorlove has 17 karma in this channel (210 overall)
[chrisaldrich], [stefp], [miklb], KartikPrabhu and ben_thatmustbeme joined the channel