#dev 2017-02-14

2017-02-14 UTC
#
gRegorLove
what is press this
#
Loqi
Press This is the WordPress bookmarklet for quickly publishing text, images, or video https://indieweb.org/Press_This
#
gRegorLove
what is bookmarklet
#
Loqi
A bookmarklet is a web browser bookmark that runs javascript commands on the current page https://indieweb.org/Bookmarklet
tantek and gk{\`_^}1wm joined the channel
#
@moncafe003
WebmentionがW3C勧告に ? 2017/01/16(29日前)???39 ?#XML-RPC 33日ぶり? ?#分散処理(15th/59) ?#IT用語_(2nd/9) https://www.mitsue.co.jp/knowledge/blog/frontend/201701/13_1201.html
(twitter.com/_/status/831354865842532352)
tantek, gRegorLove, cweiske, KevinMarks and [kevinmarks] joined the channel
#
[kevinmarks]
I should write up my svg sanitising
#
Loqi
[kevinmarks]: sebsel left you a message 9 hours, 31 minutes ago: Oh thanks! I actually already implemented a Twitter login on my site before I saw your message :)
#
Loqi
I agree
#
loqi.me
edited /Press_This (+47) "petermolnar added "https://petermolnar.net/press-this-indieweb/" to "See Also""
(view diff)
KevinMarks joined the channel
#
@franckpaul
@nhoizey webmention → tu causes chez toi et tu « link » ensuite. Mais tu peux pas commenter **chez** quelqu’un. C’est complémentaire.
(twitter.com/_/status/831434870782033920)
#
@nhoizey
@franckpaul bin si l’autre inclu ton texte dans sa liste de webmentions, c’est tout comme…
(twitter.com/_/status/831436314004291586)
#
@nhoizey
@franckpaul je suis en cours de transition (Webmentions depuis 2 semaines seulement), ET malheureusement, peu de gens utilisent Webmentions…
(twitter.com/_/status/831450886111113216)
#
@nhoizey
@franckpaul de mon côté, je réceptionne les webmentions, mais je ne les envoi pas encore, il faut que je le fasse.
(twitter.com/_/status/831451279549423616)
#
@nhoizey
@franckpaul je ne sais pas si je conserverai Disqus ou pas, pour ceux qui ne font pas de Webmentions…
(twitter.com/_/status/831451493819641856)
#
@franckpaul
@nhoizey côté Dotclear, à côté des comms, on gère les traditionnels trackbacks, pingbacks et webmentions (c’est du all inclusive quoi) ;-)
(twitter.com/_/status/831452038605242368)
sandro_ joined the channel
#
petermolnar
I found an interesting problem: I'm using a fork of https://github.com/vrypan/webmention-tools to find the author on a page with Beautifulsoup as .find(True, attrs={'class':'h-card'}) or .find(True, attrs={'class':'p-author'}). myfreeweb reposted a post of mine at https://unrelenting.technology/notes/2016-07-20-15-59-46 and the reponst contains my h-card as well, and because it's the first h-card on the page this way, that's the
#
petermolnar
h-card that gets found. Any ideas how to address this with the same find method? Basically I'm trying to figure out how to limit the search depth or to exclude elements with certain parent classes.
#
petermolnar
oh. the h-enty is on the <body> on that site. That complicates things
#
petermolnar
I think I got it, if .find_parents(True, attrs={'class':'e-content'}) finds anything, move on
#
Zegnat
You are using a full HTML parser for finding microformats? I would like to do that too, but my options in PHP seem limited
#
cweiske
what are your issues?
#
Zegnat
whose, cweiske? Mine are just that most HTML parsing seems to be done with DOMDocument in PHP, and I am not a fan of DOMDocument for HTML.
#
cweiske
yours, yes
#
voxpelli
Zegnat: there are full HTML parsers for PHP
#
cweiske
domdocument is the only way to load non-wellformed html in php
#
petermolnar
"full valid XHTML strict... sooo bauhaus :P" - @aorcsik (a friend of mine, a long while ago)
#
cweiske
my page is xhtml strict
#
cweiske
and sends application/xhtml+xml as content type
#
petermolnar
Loqi, give cweiske a bauhaus badge
#
petermolnar
what is XHTML?
#
Loqi
XHTML is a markup standard (W3C) for HTML that has been superseded by HTML5 https://indieweb.org/XHTML
#
petermolnar
what is XHTML Strict?
#
Loqi
It looks like we don't have a page for "XHTML Strict" yet. Would you like to create it?
#
petermolnar
I wasn't aware WordPress is backporting security patches to old version if it's severe enough
#
GWG
They've been doing that for some time
#
petermolnar
interesting
#
petermolnar
probably this is what saves half of the web from collapsing
#
petermolnar
sorry, 25% :)
#
petermolnar
but backporting to several major versions is not a good approach, to be honest
#
petermolnar
we're at 4.7
#
seblog.nl
edited /Private-Webmention (+610) "added myself"
(view diff)
#
petermolnar
why is 4.2 still supported?
#
GWG
People refusing to keep software up to date is not a new issue
#
petermolnar
yes, and they should learn it the hard way, if the sw is internet-facing
#
petermolnar
or networked
#
cweiske
that's why backports
#
petermolnar
(sw from the 80s on an air-gapped PLC is not a use-case here)
#
petermolnar
but that's the same mentality as build for php 5.2
#
cweiske
how I hate major updates that always break things
#
petermolnar
i haven't updated my cache plugin since 4.3? I think, still no need to
#
petermolnar
I disagree that they break things
#
cweiske
I don't talk about wordpress, I talk about updates in general
#
cweiske
especially web applications
#
petermolnar
yeah, dovecot 1.x to dovecot 2.x (IMAP server) was rude
#
cweiske
take typo3
#
petermolnar
no, thank you, cweiske, you can keep typo3 :)
#
cweiske
I still have projects at 6.2 although 7.6 is stable since > 1 year, and 8 LTS release is near
#
GWG
I would like to see WordPress push a bit more.
#
cweiske
because if you try to update, things break. developers use major version jumps to remove deprecated or undeprecated code
#
cweiske
and things break
#
GWG
But you can't push people on the scale they have without consequences
#
petermolnar
wp would benefit from an LTS release for sure, but their current way of keep things compatible and save people from themselves is just terrible
#
cweiske
"current way" is backporting, right?
#
petermolnar
even centos deprecates centos6 in 2018
#
petermolnar
wp seems to be keeping things alive even longer, which is not healthy
#
cweiske
depends on the resources you have to backport
#
cweiske
or take libraries
#
cweiske
I've heard colleagues complain about total incompatibilities between guzzle 5 and 6
#
petermolnar
cweiske reasonable compatibility, deprecation notes in advanced, sure
#
petermolnar
PHP is doing rather well under these term, 2 version of deprecation notes, for example
#
petermolnar
but it's not healthy to maintain forever backwards compatibility
#
cweiske
another example is typo3's flux/fluidcontent/vhs extensions
#
cweiske
they do major version jumps every couple of months
#
petermolnar
that if, of course, not healthy
#
cweiske
and if you update you can be sure that it breaks code
#
petermolnar
the half a year ubuntu releases are pain as well, I'm sticking to LTS
#
cweiske
my laptop is still running ubuntu 14.04
#
cweiske
because upgrading sucks
#
petermolnar
so sure, LTS is needed, proper deprecation warnings are needed, and so on, but you need to move forward at some point
#
petermolnar
I'm having so much pain with mint 18 that I'm seriously condisering stepping back
#
cweiske
in my eyes people deprecate too early and don't care about BC mostly
#
petermolnar
most of them, sure, WP is definitely not one of them :)
#
Zegnat
voxpelli: http://masterminds.github.io/html5-php/ converts to and from DOMDocument though. I just always feel a bit iffy about that. But I will check it out slightly more indepth soon
singpolyma and mbrock joined the channel
#
aaronpk
petermolnar: is there a reason you don't just parse the whole page and deal with the resulting mf2 data structure?
#
aaronpk
it might be less work overall :)
#
petermolnar
no, this ended up to be efficient enough
#
aaronpk
because that author h-card won't be the "first" one anymore, it would move to a nested object of the h-entry
#
petermolnar
that is fine; I'm doing a find_all and exclude all of those which have e-content class as parent
#
petermolnar
the first of the remaining is good enough
#
aaronpk
i just think you're going to end up with a bunch of other edge cases doing it this way
#
petermolnar
not impossible, and I'll probably consider moving back to full mf2 parsing the next time I find something like this, but this is still a fast and easy way
#
voxpelli
yeah, using a proper mf2-parser that passes all of the tests that has been created for an mf2-parser should be preferable
#
voxpelli
else there's always a risk that someone can have a special case that's supported by tested parsers but which one hasn't thought of oneself
#
petermolnar
but... I just want to find the h-card...
#
aaronpk
what is authorship?
#
Loqi
authorship is a claim about who the author(s) of a post are https://indieweb.org/authorship
#
aaronpk
take a look at that list and tell me you want to do that without using a mf2 parser :)
[cdevroe], singpolyma and tantek joined the channel
#
KevinMarks
@petermolnar also, if you're using beautiful soup, make sure it is using html5lib
#
KevinMarks
And not the dodgy built-in html parser
Jocelyn1 joined the channel
#
martymcguire[m]
oh no. quill deprecated support for the "post" scope?
#
aaronpk
yeah in favor of create/update/delete
#
aaronpk
but i left a setting there on the authorization screen for fallback support!
#
martymcguire[m]
i saw the setting for falling back to post earlier, but i don't see it now
#
aaronpk
if you log in a second time it bypasses that prompt screen and should request the same scope that was previously granted
#
aaronpk
yeah the prompt is after that page the first time you sign in
#
martymcguire[m]
hmm. i must have selected create without realizing it.
#
aaronpk
ah yeah that would do it. i guess i need a way to reset that default
#
martymcguire[m]
not sure how to get back. clearing all "application" data in chrome dev tools and logging in still leaves me with "create" scope.
#
aaronpk
yeah it's stored in quill
#
aaronpk
i'll have to make a reset button in the settings screen
#
martymcguire[m]
oh, haha
#
Loqi
awesome
#
martymcguire[m]
welp. good time for me to get my micropub endpoint supporting create!
#
aaronpk
yeah it's not a lot of code to support both :)
[keithjgrant] joined the channel
#
[keithjgrant]
What's this create mode?
#
aaronpk
just a new name for the "post" scope
snarfed joined the channel
#
[keithjgrant]
Is that part of the MicroPub spec?
#
snarfed
petermolnar: re your q earlier about finding the right h-card on a page...sounds like http://indieweb.org/authorship ? (or http://indieweb.org/representative_h-card ? )
#
aaronparecki.com
edited /scope (+65) "update scope usage"
(view diff)
#
snarfed
(petermolnar: also vrypan's repo is unmaintained. feel free to use https://github.com/snarfed/webmention-tools , it has more recent bug fixes and features)
#
snarfed
heh, sorry, should have finished read logs first :P
gRegorLove joined the channel
#
petermolnar
snarfed I might send a pull request, but I need to sit down to merge the chages; didn't know you picked the repo up
#
snarfed
well, my fork at least :P
singpoly1a joined the channel
#
voxpelli
just released update of node.js micropub-express module that treats "create" as default scope but still also supports "post" scope
#
aaronpk
voxpelli++
#
Loqi
voxpelli has 6 karma in this channel (100 overall)
#
voxpelli
yay! three digits!
#
Loqi
does a happy dance!
KartikPrabhu joined the channel
#
kodfabrik.se
edited /scope (+218) "/* Scopes accepted by IndieWeb sites */ Adding myself"
(view diff)
#
voxpelli
very very very much needs to get autolinking done for notes posted to his Micropub endpoint from eg. Quill
tantek, KartikPrabhu and [keithjgrant] joined the channel
#
[keithjgrant]
@voxpelli How is authentication done in webpage-micropub-to-github? (On inbound requests from, say, Quill)
#
voxpelli
[keithjgrant]: it is checked against indieauth.com by default, or against any other server one has configured instead
#
voxpelli
rechecked – there's no default, so it checks it against whatever MICROPUB_TOKEN_ENDPOINT you have configured
#
voxpelli
[keithjgrant]: implementation is based on the "tokenReference" option here: https://github.com/voxpelli/node-micropub-express#options
#
aaronpk
good :) I would have recommended against defaulting to indieauth.com
#
[keithjgrant]
Ah, cool. Thanks
#
[keithjgrant]
voxpelli++
#
Loqi
voxpelli has 7 karma in this channel (101 overall)
tantek joined the channel
#
aaronpk
we are now up to 4 implementations of sending private webmentions, and 3 of receiving!
#
sebsel
I plan on evening those numbers soon!
#
aaronpk
this is a good sign! none of the other attempts at private webmention had this many implementations
#
tantek
that is a good sign
[keithjgrant], tantek, [eddie] and miklb joined the channel
#
tantek
catches up on this morning's logs
#
tantek
petermolnar: why does it complicate things to put h-entry on the <body> ? that's actually a recommended practice for post permalink pages - that way you are indicating this entire page represents a post
Loqi, KevinMarks and aaronpk joined the channel
#
colintedford.com
edited /Node.js (+169) "/* Projects */ Skein"
(view diff)
#
sebsel
also wandering: I should detect authors on urls I mention in my private posts, and then add those to the audience/whitelist of people who can login.
#
sebsel
I might only want to do that for in-reply-to and like-of links though. Not sure.
#
aaronpk
i decided i didn't want to do that automatically
#
aaronpk
same behavior as mentioning people in private slack conversations. they aren't automatically added to the conversation
KartikPrabhu joined the channel
#
sebsel
hm, yeah. but that would require some webmention UI for me, then, because I can't manually generate the token and stuff.