#dev 2016-12-21

2016-12-21 UTC
KevinMarks joined the channel
#
KartikPrabhu fixed non-break space problem in fragmentioner.js by hacking all browser selections to work like FF's bad behaviour!
#
tantek.com edited /Instagram (+30) "/* Single Photo Bookmarklet */ update for new IG web front-end. They moved the caption from img alt to the first h1." (view diff)
#
tantek KartikPrabhu: what happens when FF fixes that bug?
#
KartikPrabhu if no changes are made, it will still work
#
KartikPrabhu i mean no changes to my new code
#
KartikPrabhu the new hack simply replaces all whitespace characters by a normal space in the selection text before generating the fragmention
#
tantek nice
#
KartikPrabhu this is also what fragmention.js does while comparing to find the element in the HTML
#
KevinMarks That's good, as medium does that nbsp trick too
#
KartikPrabhu oh!
#
KartikPrabhu also FF selection does this only on whitespace characters. For emdash or endash FF does no funny thing
#
KartikPrabhu or even a non-breaking-dash
#
tantek whitespace processing is hard
#
KartikPrabhu true
#
KartikPrabhu might send PR to adactio's wook for this
#
KartikPrabhu realises that he can't write marginalia on his notes since notes are plaintext represented!
#
tantek sounds like a good 2017-01-01-commitment goal!
#
tantek marginalia on notes!
#
KartikPrabhu I already have a bunch: https://kartikprabhu.com/notes/2017-indieweb-commitments
#
Loqi [Kartik Prabhu] 2017 indieweb commitments * fix responsive images height issue: currently my responsive images don’t know their heights and so cause flow when they load. This interferes with fragmention for instance. * fix hfeed2atom to display response-context ...
gRegorLove, gRegorLove_ and tantek joined the channel
#
KartikPrabhu documented fragmention non-break space fix:https://kartikprabhu.com/notes/re-non-break-spaces-fragmention-fix
#
Loqi [Kartik Prabhu] So there are a bunch of things going on with fragmentions and non-break spaces * fragmention.js replaces all whitespaces by a normal space while searching for text https://github.com/chapmanu/fragmentions/commit/05a8200296aa78de1f9ed84c4801f5d8c42b1...
#
tantek.com edited /Events (+627) "move 2016-12-14 to past, add Nürnberg for tomorrow" (view diff)
#
tantek.com edited /next-hwc (+0) "next" (view diff)
KevinMarks joined the channel
#
tantek.com edited /events/2016-12-28-homebrew-website-club (+128) "Göteborg on hiatus til 2017" (view diff)
#
kartikprabhu.com edited /fragmention (+189) "add archive link for instapaper support post" (view diff)
#
tantek.com edited /Template:IndieWebCamp (+47) "Nürnberg 2017 is planned!" (view diff)
#
kartikprabhu.com edited /fragmention (+281) "add section on UI for fragmention generation with examples" (view diff)
#
KartikPrabhu tantek KevinMarks check/edit/add ^
#
tantek.com edited /Planning (-8) "Nürnberg is planned and pending!" (view diff)
#
tantek.com edited /Main_Page (-90) "/* Homebrew Website Club */ update for next meetups" (view diff)
#
kartikprabhu.com edited /fragmention (+628) "/* Challenges */ add selection across multiple elements challenge for generating UIs" (view diff)
#
KartikPrabhu aaronpk: on indieauth log in interface it would be good to know the year of last scan too
KevinMarks_, tantek and cweiske joined the channel
#
Zegnat aaronpk: "twitter doesn't return an http 301 redirect always on t.co" - somewhere in the logs there is evidence of me banging my head for this on my rel-me validation
#
Zegnat aaronpk: the 301 vs 200 seems UA dependent
#
Zegnat And also evidence of you knowing about it ;) https://chat.indieweb.org/2016-08-02/1470179017035000
#
Loqi [aaronpk] oh it was because twitter is doing user agent sniffing on the t.co redirect
tantek, dkm_ and Zegnat joined the channel
#
marinin.xyz created /User:Marinin.xyz (+124) "Create marinin.xyz page" (view diff)
#
www.svenknebel.de edited /fragmention (+349) "/* Generating fragmention from a selection across multiple elements */" (view diff)
#
Zegnat KartikPrabhu: I don't think browsers expose element categories in DOM. But, you know what, I never checked!
#
KartikPrabhu yeah, so that could be an obstruction to implementing any element categories in javascript
#
Zegnat Well, the spec defines them simply by tag names, so it is nothing more than a list of tag names
#
KartikPrabhu right, but something like Node.category proeprty might be good?
#
Zegnat It would be Node.categories ... I am not sure how useful that will be in practice. I bet smarter people have decided that it wouldn't be useful, haha
#
Loqi hehe
#
KartikPrabhu yeah it is a weird thing so documenting and getting more eyes on it would be helpful. Hopefully you can document this stuff when you have the time
#
Zegnat IndieWeb wiki is probably not the place to document often unused HTML spec categories ;)
#
KartikPrabhu no but you could phrase the problem with suggestions about element categories
#
KartikPrabhu the problem does seem to be accesibility
#
petermolnar if there an mf2 tag to indicate language of e-content?
#
cweiske (I use lang="de" or "en" as <html> attribute)
#
petermolnar but that's for the whole pag
#
petermolnar e
#
cweiske HTML also has a dedicated attribute for single paragaphs
#
petermolnar and in case of h-feeds, they can contain varying language h-entries
#
cweiske yes.
#
cweiske "lang" is a global attribute
#
KartikPrabhu petermolnar: is there a use-case?
#
petermolnar my site? :)
#
KartikPrabhu no as in a use-case for the consumer of mf2?
#
cweiske feed-reader-should-not-display-dutch-posts
#
petermolnar I was asked why I'm not adding the lang tag to html; this is why, so I'm not certain about the other end
#
petermolnar I'd say pull in auto translate
#
petermolnar but
#
Zegnat I protest to your choice of language there, cweiske ;)
#
petermolnar that could be done without this, just with language detection
#
KartikPrabhu petermolnar: yes, I think usually mf2 properties are for consumers interpreting them
#
Zegnat Also, an h-entry with a lang: http://voxpelli.com/2015/09/oberoende-sociala-webben-2015/
#
Loqi [Pelle Wessman] Den oberoende sociala webben 2015
#
petermolnar now that's one of the weird but relevant examples, where html is en, h-entry is not
#
Zegnat Yes
#
Zegnat And as far as parsing goes, there is brain storming available: http://microformats.org/wiki/microformats2-parsing-brainstorming#Parse_language_information
#
Loqi microformats2 parsing brainstorming
#
Zegnat Lifted that brainstorming link from here: http://indieweb.org/irc/2016-05-07#t1462626687872 - the i18n session voxpelli and I did during Düsseldorf (http://indieweb.org/2016/Dusseldorf/i18n)
#
petermolnar I'm guessing that if html has this already, there's no need for and additional mf2 thing; I wasn't aware you can add lang to <whatever>
#
Zegnat It can go on anything, that's what cweiske ment when he said it is a global attribute (spec term)
#
Sebsel what is language?
#
Loqi language may refer to human (aka https://indieweb.org/language
#
Sebsel It's there too, but it can be clearer.
#
Sebsel (hm, Loqi cuts the dfn at the . after 'aka.')
#
Zegnat Yes, i18n has been on a low buffer. Probably because too few implementations are actually out there. voxpelli has one of the few
#
cweiske my posts all have a lang attribute, but my feeds miss them
#
Sebsel yeah, I don't think the parsers do anything with them.
#
Zegnat voxpelli does have lang="sv" in his feed, but not on the h-entry. On the u-url. (His feed is a list of URLs to the permalink pages.)
#
Sebsel http://pin13.net/mf2/?url=http%3A%2F%2Fvoxpelli.com%2F2015%2F09%2Foberoende-sociala-webben-2015%2F
#
Sebsel No lang there.
#
Sebsel So no lang in readers.
#
Zegnat No, it hangs on parsing brainstorming on the mf2 wiki
#
Zegnat Doesn't mean we can't start filing bugs or pullrequests to get it into the parsers
#
KartikPrabhu does anything use the lang attribute?
#
Zegnat screenreaders
#
Zegnat last time I checked
#
Zegnat Google might also use them, though they probably run their own language detection as well
#
KartikPrabhu even local lang attributes?
#
KartikPrabhu as in on elements and not global head declarations?
#
Zegnat Not sure. They do parse hreflang https://support.google.com/webmasters/answer/189077?hl=en
#
Zegnat Actually, I think I discussed hreflang with voxpelli ...
#
seblog.nl edited /language (-2) "no . in the dfn" (view diff)
#
Zegnat http://indieweb.org/irc/2016-05-07#t1462627301796
#
Zegnat KartikPrabhu, local lang attributes get picked up by screenreaders. See this article, ends with a link to a demo video: https://www.paciellogroup.com/blog/2016/06/using-the-html-lang-attribute/
#
KartikPrabhu oh ok that is a good use-case for mf2 to have a lang attribute detection
#
KartikPrabhu the question is if one should default to english
#
Zegnat For a while I made sure to have lang="nl" around my name on my h-card. So theoretically you would get a proper pronunciation from screenreaders.
#
Zegnat Not sure if I brought that over to my new design.
#
KartikPrabhu alright I'm out now... happy discussion
#
Zegnat If no language is defined, you can't default to anything. It is simply language-unknown. Unless your parser implements language detection, I guess.
#
Sebsel yeah I think just don't have a lang/language-field in the Mf2 then? there is no 'name' if there is no 'p-name' either.
#
Sebsel wait, name is a bad example.
#
Sebsel should get used to reading specs and thus learn to be precise
#
Zegnat I read specs and often still forget about things that later get specified more precisely in different sections. Specs are not always intuitive ;)
#
Zegnat sebsel: i18n / lang attributes for next HWC? ;)
#
Sebsel a topic or a to do?
#
Sebsel (because I still need to add lang="??" things to my h-entrys)
#
Zegnat I think as to-do
#
petermolnar lang="??" - wow, that looks Perl :D
#
Zegnat It was pretty fun to get checkins done (to some extend) during HWC
loicm joined the channel
#
seblog.nl edited /User:Seblog.nl (+472) (view diff)
#
Sebsel !tell aaronpk I have my own XRay running at http://xray.5eb.nl, and it does not return 500 on seblog.nl (it works!), so I can't debug any further for xray.p3k.io :( should I file an issue?
#
Loqi Ok, I'll tell them that when I see them next
#
Zegnat sebsel: https://chat.indieweb.org/dev/2016-12-16#t1481914335093000
#
Sebsel ah, so I just viewed seblog.nl via aaronpk's laptop... wow. but looking up seblog via your tunnlr-thing works too. So it's only the .p3k.io version that has a 500 on seblog and a timeout on licit.li.
j4y_funabashi joined the channel
#
sknebel sounds like some google app engine trouble then, afaik they had problems with fetching sites before if I remember right (I think that was only sites running on app engine though)
#
aaronparecki.com edited /Posts_about_the_IndieWeb (+111) "/* 2015 */" (view diff)
#
aaronpk !tell KartikPrabhu lol! I guess I never imagined someone would go >1 year without re-scanning the links!
#
Loqi Ok, I'll tell them that when I see them next
#
Loqi aaronpk: Sebsel left you a message 2 hours, 45 minutes ago: I have my own XRay running at http://xray.5eb.nl, and it does not return 500 on seblog.nl (it works!), so I can't debug any further for xray.p3k.io :( should I file an issue?
#
aaronpk hmm
#
aaronpk found the XRay logs in appengine, it's timing out
#
aaronpk which is odd because i thought i had set timeouts in curl which would prevent the whole php script from timing out
#
aaronpk it's also very hard to debug things in appengine because i have to push a whole new version up, i can't just edit files in place
#
aaronpk attempts to set CURLOPT_CONNECTTIMEOUT in addition to CURLOPT_TIMEOUT_MS
Sebsel joined the channel
#
aaronpk odd, i'm still getting a timeout error
#
aaronpk i have a feeling i know what is going on
#
aaronpk sebsel: is your server configured to serve the website on your ipv6 address?
#
Sebsel ah, that's something I wouldn't know a thing about
#
Sebsel I'll try to find out
#
aaronpk I have a sneaking suspicion that's what's going on
#
aaronpk I see an IPv6 address for your domain name, which appengine is probably using to connect
#
Sebsel hm, okay!
#
Sebsel I have found an IPv6 in my DNS yeah, but not at my server.
#
Sebsel I have a shared host, hard to find out what's going on
#
aaronpk aha okay. try removing the DNS entry then
#
Sebsel yeah, I did now.
#
aaronpk okay great. it looks like it has a long TTL, so i'll test again this evening.
#
Sebsel thanks for the time :D
#
Sebsel I'll try later too
#
aaronpk great. in any case, xray shouldn't be 500'ing on that error, so i'm trying to fix that
#
Sebsel I mailed my host too.
#
Sebsel yeah, if it's a problem on my site, it should point to me :)
#
Sebsel s/site/side
#
aaronpk i'm not sure what i can do to test that specific case
#
aaronpk it seems like it might be a thing that happens again, so would be nice to tell people
#
Sebsel I don't know much about this but: get a subdomain that points to a non-valid IPv6?
#
aaronpk oh i meant in the xray code
#
aaronpk especially because this timeout is proving difficult to capture
#
Sebsel host-guy confirms: 'Dat is een vreemde situatie.' It's strange. Server has no IPv6, but DirectAdmin thought it had. Fixed now, wait for 4 hours.
#
Sebsel But yeah, how to get an error-message out of it... no idea.
#
aaronpk awesome
#
sknebel aaronpk: what is the issue? it tries to load via ipv6 and somehow doesn't time out properly when there is no response?
#
aaronpk yeah, for some reason the curl timeout setting isn't effective, so it ends up hitting google's max execution time setting for the whole script
#
aaronpk and at that point it's too late for me to run any code, even to catch that error
#
sknebel uärgs, not nice
tantek and cweiske joined the channel
#
aaronpk oh man, the mf1 markup on readwrite.com is wonky and causing problems
#
aaronpk somehow the post permalink is ending up as the author's URL
tantek joined the channel
#
aaronpk there we go, ignoring mf1 backcompat fixed it
#
tantek wat
#
aaronpk Bad mf1 markup really messes with the mf2 conversion
#
bear curl (and the library it uses) will perform a DNS lookup for both A and AAAA and will prefer the ipv6 result
KevinMarks joined the channel
#
petermolnar a complete, clean build, with all images downsized, all content build with the static generator I'm working on: 11 minutes.
#
petermolnar It sounds horrible, but that's generating 6 sizes out of 773 jpg images and getting their exif, parsing 987 frontmattered markdown docs and 271 image meta with pandoc to html
#
petermolnar so it's not that bad
#
petermolnar apart from the image downsizing, the pandoc conversion and the exif reading with exiftool is surprisingly slow
#
petermolnar but at least stable
#
bear that's great stuff
#
petermolnar so, next step: whoosh, and getting that out of flask
#
bear you'll need flask for the bits of indieweb that require interaction - micropub and the like
#
petermolnar I know
#
bear ooh - sorry - thought you were trying to get rid of flask completely
#
petermolnar wrong wording from my side: getting it out as results for a search query
#
bear ahh
#
petermolnar btw, I still haven't figured out how to do the following: either asyncio to receive incoming, if it's a webmentions, send it to a greentlet (or similar), respond to client immediately and let the greentlet do the pulling and parsing
#
petermolnar I did the latter part, but not the first
#
petermolnar do it's still blocking, in theory
#
petermolnar but this is a problem for another day, I just get this thing up and running first
#
petermolnar The Right Way can wait :)
#
bear I punted and use redis as my queue
#
bear as I was already using it as a ephemeral store for tokens and the like
#
sknebel I use a folder on the filesystem as a queue, that maybe was *too* old-fashioned a way. but I think I prefer having the webmention verifier in it's own process I can lock down even more. it *shouldn't* be necessary, but...
#
bear I thought about using the file system but I already had redis and it has pubsub and I have that code ...
#
bear shrugs
#
sknebel I'm really not sure how one would go about doing normal web handling (wsgi) and asyncio-stuff in one
#
bear not following
#
bear not following your question
#
sknebel bear: sure, redis is a good solution for that, and saves you a lot of headaches about atomicity
#
sknebel that's what I thought petermolnar wanted to do
#
bear oh - yea
#
bear I think that's why he punted
#
bear relurks to do some house chores
#
petermolnar sknebel that is indeed the issue; unless it's not wsgi, but tornado, in standalone mode, proxied
KartikPrabhu joined the channel
#
KartikPrabhu messages Loqi?
#
Loqi KartikPrabhu: aaronpk left you a message 6 hours, 51 minutes ago: lol! I guess I never imagined someone would go >1 year without re-scanning the links!
pfefferle and tantek joined the channel
#
@call_user_func [New]pfefferle/wordpress-webmention A WebMention plugin for WordPress http://wordpress.org/plugins/webmention/ https://packagist.org/packages/pfefferle/wordpress-webmention (twitter.com/_/status/811700304089522176)
#
GWG Pfefferle has been busy
#
pfefferle GWG :)
#
GWG Hi, pfefferle...long time no chat
#
GWG I owe you a PR or two to try and get Webmentions out the door
#
pfefferle GWG yes, not that easy with the different timezones
#
pfefferle GWG no problem, I am also not that active
#
GWG pfefferle, still, nice to overlap
#
GWG pfefferle, it just feels like a good 2017 commitment to release
#
pfefferle GWG sure :)
#
GWG I may pause my location work for the few pieces I want to send.
#
pfefferle GWG shouldn't be that much work, at least for the WM plugin
#
pfefferle GWG would be nice to add the changes to the SL plugin
#
GWG I would like both too, but as an end of year thing...
#
pfefferle GWG https://github.com/indieweb/wordpress-indieweb/pull/44
#
GWG I saw, I am going to be looking in a few. I had to stop at the store
#
GWG I need to read up on composer
#
loqi.me created /rel-syndicate (+28) "prompted by tantek and dfn added by tantek" (view diff)
#
loqi.me created /syndicate-to (+99) "prompted by tantek and dfn added by tantek" (view diff)
#
pfefferle so I am off for now
#
tantek what is 2017 commitment?
#
Loqi It looks like we don't have a page for "2017 commitment" yet. Would you like to create it?
#
tantek 2017 commitment is /2017-01-01-commitments
#
Loqi ok
#
loqi.me created /2017_commitment (+35) "prompted by tantek and dfn added by tantek" (view diff)
#
tantek.com moved /fsws2010 to /FSWS2010
#
tantek.com edited /SWAT0 (+88) "cite inception date, post, event" (view diff)
#
tantek.com edited /hacker_images (+4) "/* Jesse Eisenberg */ linky" (view diff)
#
tantek.com created /The_Social_Network (+109) "stub dfn" (view diff)