#dev 2017-01-17

2017-01-17 UTC
#
@fb2k_readonly “WebmentionがW3C勧告に | フロントエンドBlog | ミツエーリンクス” https://www.mitsue.co.jp/knowledge/blog/frontend/201701/13_1201.html (twitter.com/_/status/821150418650566656)
#
aaronpk !tell jeena my parser is getting confused by your markup. is there a reason you have e-content on an element way outside where your actual text is? It's including your image alt text and such in the content this way. It would be much nicer if you had "e-content p-name" on the same element. see: https://xray.p3k.io/parse?url=https%3A%2F%2Fjeena.net%2Fphotos%2F265&pretty=true
#
Loqi Ok, I'll tell them that when I see them next
#
@inabe49 WebmentionがW3C勧告に | フロントエンドBlog | ミツエーリンクス https://www.mitsue.co.jp/knowledge/blog/frontend/201701/13_1201.html%23.WH1n_2jx_J4.twitter (twitter.com/_/status/821155211104243712)
#
@kgsi “WebmentionがW3C勧告に | フロントエンドBlog | ミツエーリンクス” https://www.mitsue.co.jp/knowledge/blog/frontend/201701/13_1201.html (twitter.com/_/status/821157685756182528)
#
@Jxck_ Trackback/Pingback をシンプルに蘇らせた標準仕様、 rel=webmention で www-form-url-encoded を送る。 / “Webmention” https://webmention.net/draft/ (twitter.com/_/status/821164003263016961)
#
@N_write_ Webmention 　　「枕狐」「無産」まとめ　　http://togetter.com/li/924722 (twitter.com/_/status/821165194776875008)
#
@Stocker_jp WebmentionがW3C勧告に | フロントエンドBlog | ミツエーリンクス https://www.mitsue.co.jp/knowledge/blog/frontend/201701/13_1201.html PingbackはSPAMだらけで使われなくなった印象だけど、これは対策されているのかな？ (twitter.com/_/status/821166019112644609)
#
@webcre8 全然知らなかった / WebmentionがW3C勧告に | フロントエンドBlog | ミツエーリンクス https://www.mitsue.co.jp/knowledge/blog/frontend/201701/13_1201.html (twitter.com/_/status/821170826489671680)
#
@Doveleahr Indie Author Answers http://www.thejugglingauthor.com/indieauth/ https://twitter.com/Doveleahr/status/821191330579775491/photo/1 (twitter.com/_/status/821191330579775491)
tantek, KevinMarks, KevinMarks_ and cweiske joined the channel
#
Zegnat aaronpk, I think your Jeena-parsing-problem is just another prime example of the question whether the photo of a photopost (marked up as u-photo) should be seen as - and included in - the post content (marked up as e-content)
pfefferle joined the channel
#
petermolnar !tell aaronpk https://xray.p3k.io/parse?url=https%3A%2F%2Fpetermolnar.net%2Fcategory%2Fphoto%2F :( - pin13 is happy though
#
Loqi Ok, I'll tell them that when I see them next
#
pfefferle good morning
#
Loqi rise and shine!
#
GWG Good morning
#
petermolnar morning
#
@masawo "WebmentionがW3C勧告に | フロントエンドBlog | ミツエーリンクス" https://www.mitsue.co.jp/knowledge/blog/frontend/201701/13_1201.html (twitter.com/_/status/821310814321803266)
#
pfefferle good morning GWG and petermolnar
#
petermolnar what's up, pfefferle?
#
GWG pfefferle: Anything new going on?
#
voxpelli Zegnat: did I hear a parsing problem? Can we get a parsing problem example into https://github.com/voxpelli/node-webmention-testpinger ? :)
#
petermolnar people who downsize images: are you using any specific filter? (not the instagram, I mean the lanczos, roboidux, bicubic, etc filters) or do you sharpen the downsized images? If yes, may I ask the settings? I'm trying to compare methods, because right now my images look a little oversharpened and I'm yet to find the balance.
#
cweiske I simply use "convert -resize" when creating thumbnails
#
cweiske but I don't really care how they look
#
cweiske convert's output was always enough for me
#
Zegnat voxpelli. Possibly. The discussion is between putting img.u-photo inside the .e-content of outside and that is still ongoing. I am not even sure if it makes sense to support both formats or if the community should settle for one.
#
Zegnat It looks like people are settling on that img.u-photo goes outside of .e-content. I think sknebel already moved them outside (to name 1).
#
voxpelli Zegnat: cool, would be nice to have an example of that then in the testpinger ;)
#
petermolnar I have a problem with that... a photo post is essentially the u-photo itself, so it should be part of e-content, in my interpretation
jonnybarnes joined the channel
#
sknebel Zegnat: I haven't actually changed anything yet, for various reasons
#
Zegnat Ah, sorry sknebel. I recalled wrong then
#
sknebel I said I'd do it at same point, but didn't yet, both for lack of time and it not seeming so clear if it is necessary
#
voxpelli seems like clients should probably be able to handle both cases
#
Zegnat Alright
#
Zegnat 4 hours until Have you tried to create some .e-content > .u-photo and .u-photo + .e-content examples for https://github.com/voxpelli/node-webmention-testpinger yet, Zegnat?
#
Loqi I added a countdown scheduled for 2017-01-17 7:02pm CET (#5982)
#
Zegnat I should be home and able to find some examples by then
#
voxpelli I guess we could look at packaging up the data of that repo as some kind of indiereader.rocks test eventually
#
Loqi yea
#
Zegnat That would just be taking every example post and concatting them into a feed though, right?
#
Zegnat Although you would need some different feeds to cover feed tests themselves (like author discovery on the feed and such)
#
voxpelli Zegnat: yeah, pretty much, and maybe working a bit with publish dates to emulate new posts appearing
#
voxpelli One could take some inspiration from https://github.com/playerfm/testdata
#
@ajspadial Test: A #webmention example. http://hoos.spadial.com/redia/2017/01/16/howto-cite-redia.html (twitter.com/_/status/821362390172336128)
#
@Inkwater_Masha The 21 Most Expensive Mistakes To Avoid When Writing Your Book http://authorunlimited.com/21-mistakes-report-pdf/?platform=hootsuite #selfpub #indieauth (twitter.com/_/status/821362760713961472)
#
loqi.me created /bedroll (+104) "prompted by Zegnat and dfn added by Zegnat" (view diff)
#
loqi.me edited /bedroll (+52) "/* See Also */ new section" (view diff)
#
cweiske what is bedroll?
#
Loqi bedroll is a blogroll of people that have stayed over at your house https://indieweb.org/bedroll
#
aaronpk good morning
#
Loqi aaronpk: petermolnar left you a message 5 hours, 9 minutes ago: https://xray.p3k.io/parse?url=https%3A%2F%2Fpetermolnar.net%2Fcategory%2Fphoto%2F :( - pin13 is happy though
#
cweiske hiho
#
Loqi guten morgen
#
aaronpk petermolnar: yeah XRay doesn't support feeds yet. XRay is more than an mf2 parser, it's aware of vocabularies, and I haven't decided how I want feeds to be parsed yet
#
petermolnar I've tried on non feeds as well
#
petermolnar same result
#
petermolnar https://xray.p3k.io/parse?url=https%3A%2F%2Fpetermolnar.net%2Fgoing-static%2F
#
aaronpk aha that's a different problem
#
aaronpk your u-url property has a double slash
#
aaronpk https://petermolnar.net//going-static
#
aaronpk so when you parse the URL https://petermolnar.net/going-static/ it doens't match that u-url property
#
aaronpk and because you have two top-level h-* objects it looks like a feed unless the URL of one of the objects matches the URL of the page it's retrieved from
#
petermolnar interesting assumption, there is a h-entry and a h-author at top level, true
#
petermolnar but why would it be a feed without h-feed?
#
aaronpk because some people publish just a list of h-entrys as a feed
#
petermolnar so 1+ h-entry => feed, ok
#
petermolnar but 1 h-entry? :)
#
aaronpk how do I know that the page that has 1 h-entry isn't just because that tag has only one post
#
aaronpk basically it comes down to: some people publish multiple h-* objects on a page. in some cases that's a post with an author like yours, in other cases it's a list of posts. in order to distinguish that case, I check for whether the URL of one of the h-* objects matches the URL I retrieved,and if so, then that page represents a single object
#
aaronpk note that if there really is only one top-level h-* object then i always return that object without checking the URL property
#
petermolnar in case of matches, would "url" and "url/" be a match or is that different?
#
aaronpk that's different
#
petermolnar I'll fix my u-urls then
#
petermolnar turned out I have to maintain / and non / as well, long story
#
aaronpk either way you have a strange double slash thing going on :)
#
petermolnar yes, I've found that already
#
petermolnar that's a typo which I've no idea how I missed
#
aaronpk oh yeah and I use the canonical URL that is redirected to when comparing those URLs
#
voxpelli there are ways to convert RSS/Atom to mf2 so maybe it doesn't have to be part of xray but rather be part of a separate service
#
aaronpk so for example it appears you redirect https://petermolnar.net/going-static to https://petermolnar.net/going-static/ which means when XRay is given https://petermolnar.net/going-static it will follow the redirect and then is assuming it will find https://petermolnar.net/going-static/ in the h-entry
#
aaronpk so you should be outputting the canonical URL in your HTML anyway
#
vanderven.se martijn edited /grav (+249) "/* Indieweb */ Add Webmention plugin." (view diff)
#
petermolnar aaronpk interesting, https://xray.p3k.io/parse?url=https%3A%2F%2Fpetermolnar.net%2Fgoing-static%2F is now parsed, despite the fact that it has a top-level h-author
#
aaronpk yes that's what i was trying to say
#
petermolnar so I fixed the url, good
#
aaronpk now that the u-url property of your h-entry matches the URL of the page that is parsed, XRay knows the page is that h-entry
#
@ringmaster @miklb Micropub? h-entry? And Webmention is just a non-XMLRPC version of Postback, with no additional benefit and all the same faults. (twitter.com/_/status/821375732412649472)
#
aaronparecki.com edited /next-hwc (+0) "Redirected page to [[events/2017-01-25-homebrew-website-club]]" (view diff)
tantek joined the channel
#
Loqi Have you tried to create some .e-content > .u-photo and .u-photo + .e-content examples for https://github.com/voxpelli/node-webmention-testpinger yet, Zegnat?
#
Loqi Countdown set by Zegnat on 2017-01-17 at 3:02pm CET
#
aaronpk lol
#
aaronpk !date
#
Loqi Now: 2017-01-17 10:03:30 PST Unixtime: 1484676210 NewCal: 2017-1-18, New Friday, the 18th day of the first bim
#
aaronpk !date CET
#
Loqi CET is 2017-01-17 09:03:36 PST
#
aaronpk no it's not
#
tantek wow those newcal dates are wrong too!
#
tantek why is it picking the 18th for them?
#
tantek when today is the 17th?
#
aaronpk huh UTC is still the 17th too huh
#
tantek also the 17th is a New Tuesday, and the 18th is a New Wednesday, so I'm not sure where it's getting "Friday" from
#
aaronpk iirc I copied a bunch of functions out of cassis a while ago for that
#
tantek I don't think I ever implemented anything for new DoTW
#
aaronpk hmm
#
tantek though it would be trivial
#
aaronpk hm i don't know where these came from actually
#
tantek array("Monday","Tuesday","Wednesday","Friday","Saturday")[dayofbim % 5]
#
tantek something like that :)
#
aaronpk that's almost exactly what i have
#
aaronpk https://gist.github.com/aaronpk/72a1ca4099e31a98d8a6e465f615542b#file-newcal-functions-php-L30
#
tantek oh 0 indexing duh
#
tantek hence it's off by one
#
tantek also a if check for 61 -> "Sunday"
#
tantek aaronpk: haha we both made the same coding error
#
Loqi rofl
#
aaronpk i don't remember writing this code, but i can't find where i would have copied it from either
#
tantek if (dayofbim == 61) return "Sunday"; else return array("Saturday","Monday","Tuesday","Wednesday","Friday")[dayofbim % 5]
#
tantek adding "new" is presentational and can be done outside that
#
KevinMarks Seems like an opportunity to pick day names that aren't gods from random pantheons
#
aaronpk is dayinbim right? 1 + (($d+1) % 61)
#
tantek KevinMarks: I think I explored that in newcal.org and determined why that was less preferable
#
aaronpk that's where 18 is coming from
#
aaronpk !date
#
Loqi Now: 2017-01-17 10:11:41 PST Unixtime: 1484676701 NewCal: 2017-1-18, New Wednesday, the 18th day of the first bim
#
aaronpk the input is php's "z", The day of the year (starting from 0)
#
aaronpk i think it's adding 1 too many times
#
tantek double checks CASSIS to make sure these aren't functions in there
#
tantek aaronpk definitely not right
#
tantek fails trivially for the first day of the year
#
tantek 0+1 % 61 = 1. then +1 = 2.
#
aaronpk ah cassis does 1+date_format('z')
#
aaronpk for ordinal days
#
aaronpk reviews http://tantek.pbworks.com/w/page/19402947/NewCalendar#YYYYDDDtoNewCalendar
#
aaronpk (DDD-1) mod 61 + 1 = dayinbim
#
aaronpk !date
#
Loqi Now: 2017-01-17 10:17:11 PST Unixtime: 1484677031 NewCal: 2017-1-17, New Tuesday, the 17th day of the first bim
#
aaronpk fixed
#
aaronpk my input was already 0-indexed, so now my function is just 1 + ($d % 61)
#
KevinMarks Moanday, Twosday, Humpday, Fryday, Satday with bonus Soonday
tantek joined the channel
#
KevinMarks Sexagesimal currency http://www.thedailymash.co.uk/news/business/baffling-decimal-currency-to-be-phased-out-20160630110021
#
KevinMarks There were 12 pennies to the shilling and 5 shillings to the half-crown
#
KevinMarks So 240 pennies in a pound, or 960 farthings
#
KevinMarks 13 shillings in a guinea, so 252 pennies or 1008 farthings
arush joined the channel
#
seblog.nl edited /Facebook (+248) "added criticism about deleted data" (view diff)
jonnybarnes, KartikPrabhu, marc___, marcthiele, marc_thiele and tantek joined the channel
#
tantek what I'm hearing about jf2 is that it was kind of a half-way approach that didn't really work well for anybody
#
tantek it simplified some things (compared to mf2 JSON) but didn't go far enough in others (authorship)
#
aaronpk hm interesting summary
#
tantek that sounds like it is not solving any use-cases particularly well, and thus perhaps it needs to be redesigned from a re-analysis of the use-cases
#
tantek with likely a renaming accordingly as well
#
tantek more focused on the use-cases than on sounding generic
#
tantek IIRC,
#
tantek "jf2" was something we settled on because of its generic sound
#
tantek KevinMarks: perhaps you could document what use case(s) you think jf2 is (should be) solving
#
KevinMarks And it does tend to spawn children
#
KevinMarks Yes, I think that's sensible. It looks like the use case is more about vocabulary convergence than remapping
#
KevinMarks The current focus is mostly on making the json more idiomatic I think
#
KartikPrabhu what's idiomatic? I here that in python-world too
#
KartikPrabhu hear*
#
KevinMarks To some extent it's making the data look less weird
#
tantek or less generic
#
tantek idiomatic like idiosyncracy is actually more weird, more specific
#
KevinMarks Eg the gdata json format looks like shit because it's trying to map xml
#
tantek that's just bad generic design, and a strawman
#
tantek anything "map xml" is a strawman
#
tantek because no one actually bothers with that in practice
#
KevinMarks No, this was the official api for a lot of products for a while. It isn't a strawman, but a design antipattern
#
tantek the antipattern is not "be generic" but rather "design your generic thing based on mapping from another generic thing"
#
tantek that's why I said it was a strawman
#
tantek as an argument against "generic"
#
KevinMarks Right. The less generic you make it, the more assumptions you are encoding. That's why this is difficult
#
tantek one of the reasons why :)
#
tantek.com edited /shared_hosting (+693) "Advantages - from IRC" (view diff)
snarfed joined the channel
#
snarfed hey micropub experts (eg aaronpk), if a micropub server receives a create or update with unrecognized properties, is it expected to still store them and return them to queries, even if it does nothing with them?
#
Loqi snarfed: ben_thatmustbeme left you a message on 2016-12-03 at 10:01pm UTC: was a test for it, but its not on the implementation reports.
#
Loqi snarfed: ben_thatmustbeme left you a message on 2016-12-03 at 10:03pm UTC: the way i am going to have inkstone image working, its storing them internally as a data URL, then when it hits the inkstone server, it checks for a media endpoint and posts them all there, if not, it will just try sending them as a data URL
#
Loqi snarfed: ben_thatmustbeme left you a message on 2016-12-03 at 10:03pm UTC: but thats just as an initial step, i'd prefer to get it working much better than that
#
Loqi snarfed: tantek left you a message 5 days, 3 hours ago: does Bridgy have its own blog or twitter handle (preferably both?)
#
Loqi snarfed: tantek left you a message 1 day, 14 hours ago: just successfully used Bridgy Publish to post an RSVP "yes" http://tantek.com/2017/015/t2/yoga-journal-market-towel-bars to the same event that the previous RSVP "interested" post got errors as discussed. Makes me think something may be going wrong on the Bridgy end (might have been the first test of Bridgy Pubilsh "interested"?)
#
snarfed https://micropub.net/draft/#unrecognized-properties implies no, but isn't explicit
#
snarfed wow whee hi loqi
#
tantek hah
#
tantek welcome back snarfed :D
#
tantek snarfed, the RSVP info is particularly interesting
#
ben_thatmustbeme woah, loqi message dump
#
snarfed heh thanks. and no, bridgy doesn't have its blog or twitter
#
tantek I haven't filed an issue yet, but can if you'd prefer it captured that way
#
aaronpk snarfed: good question. I don't think we have a good answer for that yet.
#
tantek "maybe" works, "interested" returns an error
#
tantek (tl;dr)
#
snarfed ok! thx
#
tantek only found out by actually implementing ALL of the p-rsvp values (and a UI for them) and testing them with real world content (and POSSE with Bridgy Publish)
#
tantek there's a lesson in there somewhere
#
snarfed tantek: yup. gory details in https://github.com/snarfed/bridgy/issues/717#issuecomment-273047091 . thanks for finding! off to add it to the docs now.
#
snarfed bridgy now at least reports a nicer error
#
tantek.com edited /pgp (+364) "Criticism / May not help email security - citation Fastmail" (view diff)
#
snarfed btw tantek sounds like you have github notifs (email or otherwise) turned off? ok if so, just curious
#
tantek snarfed - I think my email github notifications may get autofiled - because github sends too damn many (and/or lacks a decent configuration UI)
#
tantek this is a general problem with notifications
#
tantek really bad incentives built into the model that encourage all service providers to over-notify you
#
tantek because everyone is yelling to get your attention
#
tantek silo notifications in particular
#
snarfed definitely. tip: unfollow all repos, then you'll only get notifs for mentions and assignments, which is way more manageable
#
snarfed er, unwatch
#
tantek they don't care about your productivity, only that you visit their site and view their stuff (ads etc.) "engangement"
#
tantek !date
#
Loqi Now: 2017-01-17 21:37:26 UTC Unixtime: 1484689046 NewCal: 2017-1-17, New Tuesday, the 17th day of the first bim
#
tantek yay!
#
Loqi :D
#
ben.thatmustbe.me edited /HubZilla (+222) "/* IndieWeb Compatibility */" (view diff)
#
ben.thatmustbe.me edited /Friendica (+123) "/* IndieWeb Compatibility */" (view diff)
#
seblog.nl edited /shared_hosting (+116) "expanding after the dfn" (view diff)
#
sebsel what is ssh?
#
Loqi ssh is short for Secure SHell, a replacement for telnet https://indieweb.org/SSH
#
tantek hmm - I have ssh access on my shared hosting but I think that is a special case
#
sebsel that one can have more info too
#
sebsel Zegnat has too! I was surprised by that
#
aaronpk i do too (dreamhost)
#
tantek perhaps worth noting? sometimes has ssh access?
#
aaronpk tho by default it's disabled, but you can enable it per account there
#
tantek (could even list specific shared hosts that explicitly offer it)
#
sebsel yeah, I made a list here in my editor... just trying to think of hosts I know
#
sebsel I have 8 now, all without ssh :/
#
sebsel Maybe it's a Dutch thing.
#
sebsel even the quality host transip.nl has no ssh on their shared hosts.
#
KartikPrabhu bear will have opinions on SSH access for users :)
#
bear a proper shared host setup should allow for ssh access
#
bear most are using sftp for "ftp" access anyways - the older ftp services are rarely used
miklb joined the channel
#
sebsel I might live in another age then. I never had sftp either. it's all DirectAdmin and cPanel out here, served over http.
#
sebsel they even load jQuery via Google over http in the login-screen.
#
tantek sebsel - /web_hosting has a list of hosts in general
#
sebsel checks out
#
bear it's why I really like the DigitalOcean/Linode/Packet.net model -- you just get a small host and have root access
#
tantek !tell gRegorLove want to figure out 2017 Q1 dates for HWC meetups and create some stubs? or at least for February and then we can iterate based on actual meetings held from there?
#
Loqi Ok, I'll tell them that when I see them next
#
tantek goes to book MozSF for at least the 25th
#
tantek.com edited /events/2017-01-25-homebrew-website-club (+443) "/* Where */ SF location confirmed @MozSF! (keep usual 1st fl location in comments for future copy/pasta)" (view diff)
#
tantek !tell KevinMarks MozSF confirmed for HWC SF 2017-01-25 - go ahead and post the indie event and FB POSSE copy!
#
Loqi Ok, I'll tell them that when I see them next