#microformats 2018-03-15

2018-03-15 UTC
#
KartikPrabhu
this si why vocabulary specific tests are useless
#
KartikPrabhu
there is nothing stopping anyone from pulishing h-news
#
[tantek]
KartikPrabhu, rather, vocab tests can be useful for linters
#
Loqi
[tantek]: KartikPrabhu left you a message 25 minutes ago: here you go simplified real world example http://pin13.net/mf2-dev/?id=20180314234440988 from http://microformats.org/wiki/events/2009-06-26-microformats-4th-bday#summary. note the dtstart is not parsed
#
Loqi
[gRegorLove] I'm working on this currently. I have it working for `dt-`. For `p-` I'm wondering if this is the desired effect. It basically brings back the issue of messy implied p-name, but in the nested microformat's `value` instead. For @aaronpk's post ...
#
[tantek]
No it’s not the same
#
[tantek]
Because there’s explicit p-* markup around the whole thing
#
[tantek]
Therefore that’s author / publisher intent
[eddie], nitot and tantek joined the channel
#
@garrettc
@littlehelper @JSOxford It involves microformats, but they're just one part of it. It's more of a movement, making use of a number of specifications, to achieve a goal.
(twitter.com/_/status/974229089631461376)
Newbie1, edsu_, nitot, bear_, hober, hober2, echarlie, tantek, Garbee, barpthewire and [kevinmarks] joined the channel
#
aaronpk
i'm getting an extra Z
#
aaronpk
I don't understand where that's coming from
#
sknebel
-dev doesn't have it, so a late fix introduced it
#
sknebel
see, and php-mf2 could simply *return the string from the datetime=* instead of hunting bugs in whatever it does instead
#
aaronpk
I thought it did
#
aaronpk
oh it looks like it's mixing VCP parsing with regular parsing for that
[eddie] joined the channel
#
aaronpk
there were no tests in the test suite for a timestamp with Z
#
aaronpk
:sigh:
#
aaronpk
aha found where this is happening
#
aaronpk
it seems to be confusing the sense of an implied timezone with explicitly authored timezone
#
aaronpk
pretty sure this is only because of the VCP parsing stuff too
#
aaronpk
oh maybe not. we're in a section here with a comment: // Not using value-class (phew).
#
aaronpk
this is crazy
#
aaronpk
omg found it
#
sknebel
why does it ever try to imply timezones there?
#
aaronpk
for one very specific case
#
sknebel
ok, I wasn't aware that people started implementing that
#
aaronpk
oh is that new?
#
sknebel
not really, but not in the spec yet
#
Loqi
[tantek] #4 vcp: imply dates should also imply tz if needed
#
KartikPrabhu
yeah implied-tz is not in spec
#
KartikPrabhu
so I have left it out of mf2py
#
sknebel
but that explains my confusion why anything at all is done to those attributes
#
KartikPrabhu
can't read php!
#
aaronpk
haha that's how I feel about python
#
Loqi
rofl
#
aaronpk
technically I only changed a regex, no php
#
sknebel
assuming the overall logic around that is correct it seems like a correct fix for this specific issue
#
KartikPrabhu
aaronpk: I really can't read regex! :P
#
aaronpk
I added a test that has a datetime authored value of 2012-08-05T14:50:00Z since apparently there weren't any before which is how this was allowed to break
#
sknebel
that's going to be interesting when turning this into a spec change: what timestamp parsing does have to happen on the value
#
sknebel
I think it probably makes sense to pull dt-parsing details out of VCP then, and generalize it
#
Zegnat
I think I have looked into that before sknebel. But I couldn’t think of a good way to preserve all the things that vcp does without just copying vcp without generalisation
#
sknebel
it's pretty similar likely. well. I guess one could also co-opt the valid html5 datetime= values, but that's pretty close to restating vcp rules (I think am/pm isn't covered there)
#
Zegnat
HTML does not have am/pm in time strings, no
#
aaronpk
well if someone can comment on that PR I want to merge it in and release an update
#
sknebel
commented
#
sknebel
I'd like to enquire if YYYY-DDD dates shouldn't also be matched, but that's not part of this
#
aaronpk
good question
#
sknebel
(I can't remember seeing those in the wild, but they are a format in VCP, and html, and ISO8601)
#
KartikPrabhu
mf2py does not YYYY-DDD yet
#
Zegnat
Ha, merged while I was commenting with a description for future reviewers :P
#
Zegnat
I wish for a world where there is 1 vcp algorithm for finding a string value for a property, and a separate date-from-string algorithm that normalises any string value (whether from vcp or not) into a timestamp for dt- properties.
tantek joined the channel
#
gRegorLove
aaronpk: Sorry for the trouble with that Z
#
aaronpk
*whew* updated xray and aperture to use the new parser
#
gRegorLove
"Not using value-class (phew)" was definitely my comment, heh
#
aaronpk
only affected 23 entries in aperture, not bad
#
gRegorLove
I jumped ahead in the change control process a bit for implied TZ, sorry about that. Commented: https://github.com/microformats/microformats2-parsing/issues/4#issuecomment-373457720
#
Loqi
[gRegorLove] This is implemented in php-mf2 since [v0.4.0](https://github.com/indieweb/php-mf2/releases/tag/v0.4.0). I realize I did not first "broaden implementer consensus" per [change control](http://microformats.org/wiki/microformats2-parsing#change_control) ...
barpthewire, [asuh], [squorch], KartikPrabhu, gRegorLove_, gRegorLove__, [kevinmarks] and tantek joined the channel
#
tantek
edited /h-event (+172) "note that p-description is proposed to be replaced by e-content"
(view diff)
#
tantek
ALL: please give feedback (even if just thumbs-up) on https://github.com/microformats/h-event/issues/3
#
Loqi
[tantek] #3 Replace h-event 'description' with 'content' property
#
KartikPrabhu
there is a proposal in mf2py to move backcompat rules to JSON files just like parser tests. Here is a suggestion https://github.com/microformats/mf2py/issues/70#issuecomment-373555316 feel free to weigh in everyone
#
Loqi
[kartikprabhu] I like the idea of moving backcompat rules into JSON to be shared among parsers. I propose the following format for the files for each mf1root containing a JSON object e.g. `hentry.json` ```json { "type": ["h-entry"], "properties": { ...
tantek and [tantek] joined the channel
#
Loqi
[kartikprabhu] I like the idea of moving backcompat rules into JSON to be shared among parsers. I propose the following format for the files for each mf1root containing a JSON object e.g. `hentry.json` ```json { "type": ["h-entry"], "properties": { ...