#dev 2017-06-11

2017-06-11 UTC
[shaners] and leg joined the channel
#
loqi.me
created /open_collective (+158) "prompted by gRegorLove and dfn added by gRegorLove"
(view diff)
#
gregorlove.com
edited /open_collective (+80) "link dfn, link indieweb profile"
(view diff)
#
@ScalaWilliam
#IndieAuth is SO MUCH easier than OAuth! https://indieauth.com/developers No secret keys, etc, etc. Works against localhost!
(twitter.com/_/status/873699890467033089)
eli_oat joined the channel
#
aaronpk
!tell eli_oat your likes are missing the "u-" prefix, they just have class="like-of" but should be class="u-like-of"!
#
Loqi
Ok, I'll tell them that when I see them next
eli_oat joined the channel
#
eli_oat
ooplah! thanks so much aaronpk
#
Loqi
eli_oat: aaronpk left you a message 17 minutes ago: your likes are missing the "u-" prefix, they just have class="like-of" but should be class="u-like-of"!
#
aaronpk
i noticed cause your like of adactio's post was showing up as a mention!
#
eli_oat
thanks for noticing. Should be fixed now
[miklb] joined the channel
#
eli_oat
aaronpk++
#
Loqi
aaronpk has 55 karma in this channel (1345 overall)
tantek and AngeloGladding joined the channel
#
tantek
oh hey Zegnat
#
Loqi
tantek: Zegnat left you a message 18 hours, 44 minutes ago: gRegorLove Nürnberg uses ? to link to videos because /? is a link to /video. Made sense to me to keep the iconography the same. I did those links, and the rest of the schedule page during IWC lunch as I was remote.
#
tantek
ok so that's my fault :P
#
tantek
that's what I get for being fast & loose with creating emoji redirects
#
tantek
alright then since it was my fault in the first place perhaps I can fix it
#
tantek
what is SnowCamp?
#
Loqi
It looks like we don't have a page for "SnowCamp" yet. Would you like to create it?
#
@ScalaWilliam
#IndieAuth is SO MUCH easier than OAuth! https://indieauth.com/developers No secret keys, etc, etc. Works against localhost!
(twitter.com/_/status/873699890467033089)
#
aaronpk
yes! haha
#
aaronpk
that's great
#
Loqi
hehe
tantek and gRegorLove joined the channel
#
gregorlove.com
edited /IndieWebCamps (+819) "+Düsseldorf, Nuremberg"
(view diff)
#
gregorlove.com
edited /Planning (-304) "/* Planned */ rm past IWCs"
(view diff)
#
wagle
eh darn, I wuz too lazy to finish making my mods to sempress a child theme.. NOW I pay! boohoohoo
tantek joined the channel
#
wagle
(bwoohoohoo?)
#
Loqi
giggles
#
wagle
actually, I don't get it.. sempress allegedly had an update, but noting happened when I updated?
tantek joined the channel
#
wagle
conclusion: my mods a couple weeks ago caused it to only now think an update to sempress theme was needed?
#
gregorlove.com
edited /2017/Düsseldorf/Schedule () "(-568) /* Saturday May 13, 2017 */ remove etherpad links after confirming they're dumped+linked on the session pages."
(view diff)
AngeloGladding, [kevinmarks] and sebsel joined the channel
#
@WendyandCharles
ReadersGazette: RT iansuth: Get Iain Rob Wright's A to Z course on self publishing. 3 videos for FREE! https://www.azofselfpublishing.com/?ref=5&campaign=Twitter #Indieauth…
(twitter.com/_/status/873868267580006403)
KevinMarks and eli_oat joined the channel
#
aaronparecki.com
edited /2017 (+29) "/* Sponsors */"
(view diff)
#
Zegnat
aaronpk, you in?
#
Zegnat
sebsel and I are banging our heads against your emoji-detector-php
#
Zegnat
Mostly we are running into trouble with the regex
#
aaronpk
it's kind of a beast
#
Zegnat
Apparently your current version does not pick-up on the rainbow flag. So we updated the map.json (needed a small tweak to the generator script) and the regex (from the latest emojione lib).
#
aaronpk
oh funny
#
Zegnat
But then it fails one of your tests ...
#
Zegnat
Apparently they changed something about the regex.
#
Zegnat
Something being: too much.
#
aaronpk
well that's strange
#
Zegnat
sebsel came up with some extra tests, and has the new regex here: https://gist.github.com/sebsel/8133fd6127d67145017d682c31c69331
#
aaronpk
awesome. which does it fail?
#
Zegnat
You will need this change to the generator or else stuff starts to fail (emoji data has a new format): https://gist.github.com/Zegnat/ef94cf0a24abf5feb34b15f31845c2e2
#
Zegnat
EmojiDetectTest::testDetectMultipleEmoji fails
#
Zegnat
The woman+heart gets seen as 1 emoji instead of 2 separate
#
aaronpk
strange, github renders it as 2 emoji as well
#
Zegnat
It should be 2.
#
aaronpk
oh right
#
Zegnat
I think the new regex might be broken, it has both the woman and heart in there. Probably part of the whole family-jwz-parsing-thing? Might just have a questionmark too much somewhere.
#
Zegnat
But I couldn’t figure it out.
#
aaronpk
yeah me either, that's why i just copied it from emojione :P
#
Zegnat
I couldn’t find a much better regex somewhere else either. Several projects seem to just be concatting the list. I thought about that as a solution, but wanted to run by you first.
#
aaronpk
making a giant list of all emoji??
#
Zegnat
basically doing implode('|', array_keys($map)) (with a little more code, obviously)
#
aaronpk
oh gosh
#
Zegnat
Could run that together with map.json generation, create a regex.json (or whatever), so you would not have a magic regex to depend on
#
aaronpk
yeah i guess that might be a more sustainable solution
#
Zegnat
Could probably do something tree-based to compact the regex. But I am not sure how we could easily do smart compression of ranges.
#
Zegnat
would like to know how emojione made their regex
#
aaronpk
ranges are easy in regex, that's how the emojione works
[kevinmarks] joined the channel
#
[kevinmarks]
It might be easier to code as an if tree
#
Zegnat
yeah, ranges are easy, but how would I easiest create them generically from our list of valid emoji? (array_keys($map))
#
aaronpk
sort by hex value and look for the contiguous groups?
#
[kevinmarks]
Can you write a regex like [☺-⚕️]
#
Zegnat
[kevinmarks]: yes, if your regex lib accept literal unicode. Else you’d want to escape those.
#
Zegnat
[rangeofsingles]|[rangeofsinglesthatallowforms]FE0F|[rangeofsinglesthatallowskinmod][1F3FB-1F3FF] ... that probably covers a lot
#
Zegnat
I am more quickly stumped when thinking about how I would automate that. I might be able to write it faster by hand, haha.
#
Loqi
hahahaha
tantek joined the channel
#
sknebel
(It's also interesting that everyone uses this emojione dataset, and not some extract from the standards. Only because the emojione thing has additional data, or is there no good authorative source that can be parsed?)
#
Zegnat
I think emojione actually compiles from the Unicode data.
#
Zegnat
Sorry, not emojione, the PHP emoji-data that aaronpk uses
#
Zegnat
Only the emojione regex was being used
#
sknebel
ah, ok
#
aaronpk
The best source of emoji data I've found is the slack one
#
Zegnat
PR sent aaronpk. Dumbest regex possible, I’ll think about something more fancy.
#
Zegnat
dumb regex does pass all tests and detects Emoji 4.0 correctly
#
aaronpk
Wow cool okay
#
aaronpk
hm tests failed because the regex.json is missing?
#
Zegnat
Did I make a mistake?
#
aaronpk
oops yeah haha
#
Loqi
rofl
#
aaronpk
it's looking for "regex.json" but the file is named "regexp.json"
#
Zegnat
ah, woops
#
Zegnat
regexp.json is right, that matches the variable
eli_oat joined the channel
#
Zegnat
I can fix before you pull
#
aaronpk
go for it
#
aaronpk
might be worth opening an issue on emojione about the improperly combined emoji i have that test for
#
Zegnat
I couldn’t get their PHP lib to work at all for me, so I couldn’t test either
#
[kevinmarks]
I wonder which executes faster, a dumb regexp or a 'smarter' one that no-one can read
#
Zegnat
PR should be fixed, aaronpk
#
aaronpk
cool, will wait for tests to pass
#
aaronpk
failed in php7
#
aaronpk
oh, this is the phpunit thing
#
Zegnat
Yeah, 7 and 7.1 fail on PHPUnit
#
aaronpk
i had to fix this for another library
#
Zegnat
[kevinmarks], I know too little about regex parsers to know about their possible performance things
#
aaronpk
we could benchmark the previous regex vs the new one
#
Zegnat
You’d have to do that with the old map JSON, as the previous regex is broken now. Saying that a broken thing is faster than a fixed thing is useless.
#
aaronpk
well it would still be an interesting comparison
#
aaronpk
because if the fancy regex is a lot faster, then it would mean it'd be worth updating the new regex to that style
#
aaronpk
ugh, gotta boot up my phpenv VM to test this
KevinMarks and sebsel joined the channel
#
tantek
what is phpenv
#
Loqi
It looks like we don't have a page for "phpenv" yet. Would you like to create it?
#
tantek
what is a VM
#
Loqi
A Virtual Private Server (AKA VPS) is a level of webhosting service where you get root access to a virtual system and can install whatever you want https://indieweb.org/VM
#
aaronpk
phpenv is a tool to help you install and run multiple versions of PHP on a single computer
#
loqi.me
created /phpenv (+118) "prompted by tantek and dfn added by aaronpk"
(view diff)
#
sknebel
aaronpk: could you add a link to that, given that there seem to be multiple different projects with that name, and a bunch of forks due to inactivity?
#
aaronpk
huh i wonder which i'm using
#
aaronpk
Zegnat: KevinMarks: quick benchmark shows that the new regex is almost twice as slow
#
aaronpk
3 seconds for 10000 `detect_emoji` calls vs 1.5 seconds
#
Zegnat
definitely need some work to add ranges back in then
#
aaronpk
speeds up a little using the 'S' modifier
#
aaronpk
ah php is already caching the compiled regex internally too
#
Zegnat
there isn’t such a thing as a regex object in PHP though, I believe. Would have liked to cache that instead of just a string in the class.
#
aaronpk
that's what i mean, php is caching it internally
#
Zegnat
after the first preg_match or something? Good on their end
#
aaronpk
yeah, it keeps a hash of the regex string used mapped to the internal compiled regex
#
sknebel
heh, even hashing a many kb string just to check if it's already cached probably wastes some time. Gotta try if that is a case where explicitly making a regex object in python makes a difference
#
sknebel
(often it doesn't, since python does this kind of caching as well)
#
aaronpk
lol that is a 57kb string wow
#
sknebel
the regex is structured with the longest sequences first as to match the multi-character combinations before the individual character they start with?
#
Zegnat
well spotted
#
Zegnat
I explicitly sort on string length before concat'ing
#
tantek
GWG, in https://github.com/Automattic/_s/issues/1113 why does he say " As h-entry isn't a one for one replacement of hentry" - it's literallly a one for one replacement.
#
Loqi
[davidakennedy] #1113 Add Schema and/or Microformats 2 Support?
#
aaronpk
something something styling of mf1 classes?
#
tantek
(also glad to see your first comment corrected that :) )
#
tantek
GWG, one minor correction: " But Core applies hentry to all post types when pages or any other hierarchical​ type would not apply as it applies to date based content." <-- every page has an author and was published originally on some date, and has a last modified, whether it is a blog post or not. Thus it can be marked up with h-entry
#
tantek
e.g. W3C specs marked up with h-entry
#
tantek
GWG, keep it up, you're doing great
#
GWG
tantek: I was taking it from the notation that it is episodic and syndicated content. But I would defer to you on that.
AngeloGladding joined the channel
#
tantek
GWG indeed it usually is, especially in the context of an h-feed
#
tantek
however it is still useful to markup any page that has an explicit author and publication date
#
tantek
which turns out is a lot of (most?) content pages
#
tantek
that could probably be better written up in a tutorial
#
GWG
The problem with pages in WordPress is it has a date, but it isn't usually displayed
#
tantek
what is u-featured
#
Loqi
featured is a proposed mf2 property (typically as u-featured) for h-entry that indicates a representative image for a primarily textual post https://indieweb.org/u-featured
#
tantek
GWG, do WordPress pages usually have an author?
#
GWG
Not in _s, the theme under discussion
#
GWG
As they all use the same data structure, they technically have one, but it isn't usually displayed. Same as the date.
#
GWG
Some themes might, but usually not
#
tantek
might be better to leave that to the theme
#
tantek
invisible data is usually a bad source
#
tantek.com
edited /h-entry (+344) "Why publish - any page"
(view diff)
#
GWG
According to the WordPress Theme Developer guide, pages are non-time dependent and without a time stamp.;
#
tantek
GWG -> see https://indieweb.org/h-entry#Why_publish and let me know if that makes more sense, especially the "Any page" para/list I just added.
#
tantek
GWG, that doesn't make sense, since everything (even pages) have to be created at some point in time, by someone
#
GWG
Yes. I was more explaining why I came to my conclusion.
[afragen] joined the channel
#
aaronparecki.com
edited /2017/Leaders (+135) "/* Discussion Channels */"
(view diff)
#
aaronparecki.com
edited /2017/Leaders (+512) "/* Discussion Channels */"
(view diff)
[chrisaldrich] joined the channel
#
sebsel
I don't know if I'm qualified to add to that page, but when thinking about a 'new channel for wiki edits' and a new channel for meta, I think those would combine well. Better than #indieweb-dev, because editing the wiki is not really dev.
KevinMarks and [miklb] joined the channel
#
[miklb]
sebsel I think if you participate in conversations in IRC/Slack you are qualified to add your opinion to give context when they meet in Portland.
#
aaronpk
go for it
#
seblog.nl
edited /2017/Leaders (+208) "/* Discussion Channels */"
(view diff)
KartikPrabhu and barpthewire joined the channel
#
@shahw1
Via JFbookman Join me at the BookBaby Independent Authors Conference—Early Bird registration open now: http://www.bookbaby.com/conference2017/ #IndieAuth…
(twitter.com/_/status/873969429201252352)
tantek, KevinMarks, [miklb] and [kevinmarks] joined the channel
#
jonnybarnes
I got my site to support ownyourgram.com
#
jonnybarnes
I think, currently I think it’ll break if the json request sends an actual h-card
#
jonnybarnes
the only post so far sent location => ['geo://xxx,yyy']
#
Zegnat
This seems interesting, https://segment.com/blog/a-brief-history-of-the-uuid/ ... I could have random unique IDs for my posts still, but they can easily be sorted on datetime
[kevinmarks] joined the channel
#
[kevinmarks]
Uuids containing sortable dates seems smart
[eddie], dougbeal|iOS and [miklb] joined the channel
#
GWG
Just sent a PR for a homepage mention setting for the WordPress Webmention plugin
[eddie] joined the channel
#
Loqi
gwg has 7 karma in this channel (230 overall)
#
[eddie]
That's awesome
KevinMarks, tantek and KevinMarks_ joined the channel
#
GWG
And it just got merged