#dev 2017-05-22

2017-05-22 UTC
tantek, miklb, [jeremycherfas], gRegorLove, cweiske and [eddie] joined the channel
#
[eddie] !tell schmarty Something weird I’ve noticed with HuffDuffer for podcasts. It doesn’t ever link back to the original post: https://huffduffer.com/schmarty/407720
#
Loqi Ok, I'll tell them that when I see them next
#
Loqi [schmarty] This Week in the IndieWeb Audio Edition • May 6th - 12th, 2017
#
[eddie] When you look at that page there is no link to https://martymcgui.re/2017/05/13/131726/
#
Loqi [Marty McGuire] This Week in the IndieWeb Audio Edition • May 6th - 12th, 2017
#
[eddie] This is unfortunate. While I’ve been working on grabbing Podcast info from web pages for some semi-automation on my ‘u-listen-of’ posts, If I start at an overcast URL (which is what I get from the overcast player, I can use the permalink to get to the huffduffer page, but then there is no way for me to get back to your original page ?
tantek joined the channel
#
[eddie] This is unfortunate primarily because unless I missed something, that’s the only podcast feed you provide is through huffduffer, correct?
[sebsel] joined the channel
#
[sebsel] Zegnat++ for working at the Kirby Toolkit Link-header fix instead of sleeping :D
#
Loqi zegnat has 5 karma in this channel (48 overall)
#
[sebsel] (still hope you got some sleep)
#
Zegnat got a little sleep
sebsel, loicm, j4y_funabashi and [kevinmarks] joined the channel
#
aaronpk i can't figure out this swarm emoji bug
#
aaronpk i *think* the foursquare api is giving me unicode code point offsets rather than emoji character offsets
#
aaronpk so for this JSON string "\ud83c\udf78 \u2014 with Joschi, Amber" (? — with Joschi, Amber) it says Joschi's name is from [10,16] and Amber is from [18,23]
#
aaronpk i might have to make some test checkins with more ZWJ emoji to really be sure
#
schmarty !tell [eddie] thanks for the heads-up. I can add "originally posted" links to my huffduffer copies in the future.
#
Loqi Ok, I'll tell them that when I see them next
#
Loqi schmarty: [eddie] left you a message 4 hours ago: Something weird I’ve noticed with HuffDuffer for podcasts. It doesn’t ever link back to the original post: https://huffduffer.com/schmarty/407720
#
aaronpk confirmed, it's definitely telling me code points
#
aaronpk so "\ud83d\udc69\ud83c\udffc\u200d\ud83c\udf73 \u2014 with Asha" (??‍? — with Asha) says the name starts at offset 15
#
Zegnat I would not be surprised if many string parsers are very happy about that, aaronpk
#
aaronpk i can't figure out how to find the right spot!
#
aaronpk preg_split in unicode mode is too "smart"
#
aaronpk hmm maybe i can totally hack this
#
cweiske aaronpk, didn't you say your server will have the iwc-video after livestreaming?
#
Zegnat aaronpk, mb_substr can’t cut in the right place using some alternative encoding? Forced 8bit encoding?
#
aaronpk cweiske: i recorded the last demo sessions to my laptop, but otherwise they're all up on youtube
#
cweiske I thought I had heard that first your blog would show the live stream, and when that's over and the server finished converting the video, it would be replaced with a video post
#
aaronpk yeah, i have that mostly working, but i used youtube this weekend instead
#
cweiske ok
#
www.svenknebel.de edited /non-interactive_IndieAuth (+637) "expand pWM example" (view diff)
#
aaronpk WELL this is super hacky but it works
#
aaronpk json_encode the string, then regex split on /(\\\u[a-h0-9]>{4}|\\\"|.)/ to turn it into an array of each "character" according to foursquare's count
#
aaronpk then the offsets i get are useful
#
aaronpk sometimes i wish i understood character encodings better
#
cweiske that sounds awful
#
aaronpk pretty much
#
aaronpk if anyone has better suggestions i'm all ears: https://gist.github.com/aaronpk/1ef5492ec24a263db65c3f36d96fc5ac
#
@NelioSoft_ES Trackbacks, pingbacks y webmentions https://neliosoftware.com/es/blog/trackbacks-pingbacks-y-webmentions/ (twitter.com/_/status/866616612174139392)
#
aaronpk in the mean time, i'm launching this
#
Zegnat launch early, iterate often :)
#
aaronpk whee
#
sknebel I'm looking into non-interactive IndieAuth right now
#
sknebel primary result for now: identity gets a lot more murky
#
aaronpk i was starting to do the same :)
#
aaronpk i think this technique can be adapted https://www.ietf.org/archive/id/draft-prodromou-dialback-00.txt
#
aaronpk i might try prototyping that on the plane
#
loqi.me created /Yahoo_Pipes (+127) "prompted by sknebel and dfn added by sknebel" (view diff)
#
sknebel I feel like a big thing with private webmention is that there is no clear identity of the receiver
#
sknebel which I think clouds my thinking here right now. and is completely different from e.g. a wiki bot
#
Zegnat aaronpk: I just ran mb_substr($input, $offset['indices'][0], $offset['indices'][1], $encoding) for every mb_list_encodings(). It is easy to extract Asha’s name in GB18030, CP936, BIG-5, CP950, and UHC. If you are looking for an even more horrendous hack.
#
sknebel wtf
#
cweiske Zegnat++
#
Loqi zegnat has 6 karma in this channel (49 overall)
#
sknebel wouldn't one expect something called "mb_substr" to work on codepoints, when the correct encoding is given? or am I missing a distinction again?
#
Zegnat Code used: https://gist.github.com/Zegnat/8dc56c2b9dc0750ec773209993b4f42a
#
Zegnat sknebel, mb_* is “too clever”. It understands that some codepoints come together to form a single character. So that’s how it will count them when you give the proper encoding. But Swarm apparently doesn’t count characters for their indices. So you need a way to access the codepoints behind the characters.
#
Zegnat If that makes sense (based on my understanding)
#
sknebel yes it does, thx
#
cweiske this is interesting: http://stackoverflow.com/a/33042226/282601
#
cweiske convert to utf-32, so every character *is* a single unicode codepoint.
#
sknebel ... that's so incredibly obvious
#
Zegnat That’s clever. My little script didn’t do conversion it just applied.
#
Zegnat is going to add another column to the table
pfefferle joined the channel
#
pfefferle good morning
#
Zegnat Most of the converted multibyte strings do what you would expect. Give you the letter “a” as that is the character on the Swarm specified offset. Note, most. EUC is interesting for returning the last two characters: http://zegnat.net/media/aaronpk.php
#
sknebel what is audience?
#
Loqi audience is an experimental property of a post that indicates the intended recipients (readers) of the post https://indieweb.org/audience
#
petermolnar I'm starting to wonder how one would store utf-32 in mysql...
#
aaronpk blob
#
sknebel for the use case here you also wouldn't necessarily have to store it that way
#
sknebel (aaronpk, you saw the link cweiske posted above?)
#
aaronpk Yeah I store it as utf8mb4 and it's fine, I just need to transform it temporarily
#
aaronpk I'll check if converting to utf32 solves it
#
petermolnar well, there is utf32 in mysql
#
Zegnat Not for the Swarm case, aaronpk. See http://zegnat.net/media/aaronpk.php
#
petermolnar I'm just curious if it needs hacks like utf8 vs utf8mb4
#
Zegnat utf32 is still going to make the emoji 1 char
#
aaronpk Well my solution works fine even if it's ridiculously hacky, so I'll just leave it alone
#
Zegnat For all you know, your hacky solution is how Swarm gets the indices in the first place
sebsel, tantek, loicm, [kevinmarks], KevinMarks, KevinMarks_, miklb, j4y_funabashi and Ruxton joined the channel
#
@eli_oat a post Voxpelli’s node-webmention-testpinger is perfect! Thank you for all the recommendations! https://mmmicro.eli.li/2017/05/22/366/ (twitter.com/_/status/866677495743090688)
[colinwalker], miklb, KevinMarks, j4y_funabashi, tantek, [chrisaldrich] and KartikPrabhu joined the channel
#
www.boffosocko.com edited /micro.blog (+343) "Mac desktop client" (view diff)
KevinMarks joined the channel
#
tantek hmm - I seem to be having a problem with Bridgy publish today
KevinMarks_, KevinMarks, gRegorLove and [cleverdevil] joined the channel
#
tantek.com edited /Events (+2) "move recent May events to past" (view diff)
sebsel, KevinMarks and KevinMarks_ joined the channel
#
tantek trying again with more debugging (Bridgy Publish that is)
#
gRegorLove Another XML feed criticism... when the GUIDs change and all of the old item show up as new again. Particularly frustrating with podcast feeds when your iPod tries to download all of them again. :)
#
gRegorLove "G"UIDs
#
tantek gRegorLove: can you document that with screenshots and specifics
#
tantek gRegorLove: globally unique != globally persistent
#
gRegorLove Such, on /RSS?
#
gRegorLove *Sure
#
tantek the problem you are seeing is not a failure of GUID, but rather a failure of impermanence
#
tantek yes, document the failure on the page for whatever format is exhibiting the failure
#
gRegorLove RIght, poor wording on my part
#
www.funwhilelost.com edited /site-deaths (+152) "/* Upcoming */" (view diff)
#
tantek the folks assuming "IDs fix all the things!" will not change their faith about that until there are sufficient counter-examples
#
gregorlove.com edited /RSS (+553) "/* Issues */ === Items appearing as new again === + screenshot placeholder" (view diff)
#
tantek well this helps a bit (in my Bridgy debugging)
#
tantek "Bridgy Publish photo to Twitter http code:400 { "error": "Publish is not enabled for your account. Please visit https://brid.gy and sign up!" }"
#
tantek I have not touched my Bridgy config
#
tantek so I wonder if somehow their tokens expired?
KevinMarks joined the channel
#
miklb have you published since you got locked out?
KevinMarks_, KevinMarks and [cleverdevil] joined the channel
#
tantek apparently: https://brid.gy/twitter/t "This account was disconnected on Twitter's end. Click to reconnect it!"
#
Loqi t
#
KartikPrabhu tantek: related to your account lock?
#
tantek possibly? ah I have not tried to Bridgy Publish to Twitter since then
#
tantek I've been using my own Twitter POSSE code
#
KartikPrabhu gRegorLove: I have had that trouble with my legacy feed subscriptions. Couldn't track the cause though
#
loqi.me created /News_Explorer (+70) "prompted by tantek and dfn added by miklb" (view diff)
#
gRegorLove I'm mostly making an educated guess there. In this instance I haven't been paying attention to the GUIDs in the feed, but I've definitely noticed it with friends switching their site from one CMS to another
#
gRegorLove And I've caused it to happen with my own Atom feed before too
#
tantek re-enables Bridgy
#
tantek hmm where should I document this
#
tantek ah yes, unlock your account
#
tantek.com edited /Twitter (+210) "/* How to unlock your account */ note AFTER steps to re enable Bridgy" (view diff)
#
tantek ok, re-auth'd and documented
#
tantek (really hoping all this documenting of experience / breadcrumbs helps others if they get similarly stuck)
#
gregorlove.com uploaded /Special:Log/upload "uploaded "[[File:RSS-old-is-new-again-2017-05-22.png]]""
#
gregorlove.com edited /RSS (+87) "/* Items appearing as new again */ screenshot sizing, caption" (view diff)
#
tantek yay that fixed it!
#
tantek Bridgy Publish working again
#
tantek (for me and my site)
#
Loqi woot
#
gregorlove.com created /Overcast.fm (+106) "stub" (view diff)
#
gregorlove.com edited /podcatcher (-16) "/* Examples */ Overcast.fm" (view diff)
#
gregorlove.com created /Overcast (+25) "r" (view diff)
[sebsel] and [colinwalker] joined the channel
#
gregorlove.com edited /jsonfeed (+163) "dfn, see also" (view diff)
tantek and [jeremycherfas] joined the channel
#
[jeremycherfas] gregorlove I had that myself just this morning, with a podcast episode from a couple of years ago suddenly downloaded again. But just one.
#
gRegorLove Huh, interesting
#
[jeremycherfas] Overcast, Show Me Your Mic.
KartikPrabhu and tantek joined the channel
#
loqi.me edited /JSON (+36) "Zegnat added "http://seriot.ch/parsing_json.php" to "See Also"" (view diff)
KevinMarks and tantek joined the channel