#indiewebcamp 2014-12-13

2014-12-13 UTC
KartikPrabhu, wolftune, mlncn-agaric and danlyke_ joined the channel
#
@kevinmarks
@pamelafox also a good question to ask in #indiewebcamp irc
(twitter.com/_/status/543586357138817025)
#
aaronpk
!tell snarfed there's some funny markup happening on bridgy posts with <p class="u-mention"> ... what happened?
#
Loqi
Ok, I'll tell them that when I see them next
#
KartikPrabhu
aaronpk: example link?
#
aaronpk
it's when my post links to other pages
#
aaronpk
i think he's trying to make the parsed mf2 contain u-mention links to the other URLs
#
aaronpk
whereas the bridgy post is a like of my original post
#
aaronpk
okay brace yourselves. webmention.io is getting some updates
mdik joined the channel
#
KartikPrabhu
i think bridgy uses it to distinguish mentions from likes and so on, but they need not be explicit
#
aaronpk
no it's good that it's explicit. it's just not valid mf2 right now
#
aaronpk
u-mention doesn't make sense on a <p> tag
mlncn-agaric joined the channel
#
aaronpk
okay that stuff is live
#
aaronpk
now we just have to wait for someone to link to the wiki
colintedford joined the channel
#
aaronpk
oh hey there is one
#
aaronpk
heh the bridgy twitter likes are different from the FB ones. it has u-like-of of both URLs https://brid-gy.appspot.com/like/twitter/xtof_fr/543410104641413121/14447132
#
aaronpk
that notification should have said "favorited a tweet linking to ..."
#
colintedford
Anyone else having trouble with the IWC front page?
#
colintedford
Like "301 Moved Permanently" trouble?
#
colintedford
!tell aaronpk I see a 301 error page where IWC's homepage should be. Other pages seem fine.
#
Loqi
Ok, I'll tell them that when I see them next
#
aaronpk
Dammit
#
Loqi
aaronpk: colintedford left you a message 38 seconds ago: I see a 301 error page where IWC's homepage should be. Other pages seem fine.
benjamin-agaric and becker_11 joined the channel
#
aaronpk
ooooops
#
aaronpk
wtf nginx
#
aaronpk
oh my bad
#
aaronpk
fixed!
#
aaronpk
thanks colintedford
#
aaronpk
that was one of the "side effects" i was worried about: https://github.com/indieweb/wiki/issues/17#issuecomment-66861083
#
colintedford
Ah, well on the bright side now you've solved *2* problems ;)
#
aaronpk
now time to solve mooooore!
#
aaronpk
running into some issues where it's going to be complicated to check permissions of all the targets specified
#
aaronpk
I wonder if I should just drop the whole private access thing and allow public retrieval of all mentions
#
aaronpk
i'm pretty sure everyone who is actualy using it is allowing public access so JS widgets can fetch the mentions
cmhobbs, snarfed and reedstrm joined the channel
#
aaronpk
okay yep.
#
snarfed
yay pretty notifs!
#
snarfed
triaging issues now
#
Loqi
snarfed: aaronpk left you a message 1 hour, 46 minutes ago: there's some funny markup happening on bridgy posts with <p class="u-mention"> ... what happened?
#
snarfed
(lurkers here, feel free to follow along on https://github.com/aaronpk/webmention.io/issues/34)
#
aaronpk
guess i should have filed that on bridgy?
#
snarfed
aaronpk: i don't quite follow your last comment on that bug
#
snarfed
thanks, looking
#
aaronpk
the first test mention I sent myself on my dev copy said "snarfed liked a post linking to aaronparecki.com/...." but really it was a like of that post
#
snarfed
ah, for non-bridgy wms
#
snarfed
i think we still need the "linking to" for bridgy wms, but just for them
#
aaronpk
no, it's not a special case for bridgy
#
aaronpk
so problem 1 is bridgy's twitter links are mis-attributing the like.
#
aaronpk
s/links/likes
#
Loqi
aaronpk meant to say: so problem 1 is bridgy's twitter likes are mis-attributing the like.
#
snarfed
debatable - we've debated it before - but ok
#
aaronpk
no, you've already switched it on facebook likes
#
snarfed
it's not silo specific. i drop links to u-mention when i've found a syndication link on your site
#
snarfed
but if i haven't, i'm inferring that links are original post links. which is obviously not ideal, but it's the compromise i've gone with so far
#
aaronpk
okay so this one is great, assuming the u-mention markup is fixed https://brid-gy.appspot.com/like/facebook/11500459/10101739853306936/501544848
#
snarfed
the question is, in the plain text of a silo post, how to distinguish original post links from mention links
#
snarfed
we've discussed ideas before, e.g. https://indiewebcamp.com/original-post-discovery#POSSE_copy_domain_approximation , but afaik we haven't found a silver bullet yet
#
snarfed
i'll fix u-mention now
#
aaronpk
are there examples of this happening on non-bridgy urls?
#
snarfed
"this" ?
#
aaronpk
"X liked a post that linked to Y"
#
snarfed
i don't think so. wmio wouldn't get the mention for that
#
snarfed
unless it happened to also handle the post's wms, which would be just a coincidence
#
aaronpk
it would if the "like" post actually showed the reply context
#
snarfed
ah. sure! but that's site-specific, since reply context display is individual preference
#
aaronpk
i want to get this right for the non-bridgy case, then we can reverse engineer what bridgy should be doing
#
aaronpk
yeah but if wmio gets a webmention for a like post that has reply context, it's going to need to craft that sentence
#
snarfed
right. i guess i just expect that will be rare to never
#
aaronpk
really? i think that is quite likely
#
aaronpk
my site would do it if i published likes
#
snarfed
regardless, i understand what you're going for. i'm all for it. :P
#
aaronpk
it should do it for replies now, i'm just trying to find an examples
#
aaronpk
here's one, not a very good example, but still: https://aaronparecki.com/replies/2014/11/27/1/indieweb
#
aaronpk
that should result in a webmention sent to wmio with a target of indiewebcamp.com as well as one for the /wiki/backup page
#
snarfed
yup, agreed
#
aaronpk
the text shoudl be "aaron replied to a post that linked to http://indiewebcamp.com"
dns53 joined the channel
#
snarfed
ok. so if we see the source is a reply/like/etc (based on mf2), but the in-reply-to (etc) link *isn't* the target, then we do this code path of "aaron Xed a post that linked to…"
#
aaronpk
that sounds reasonable
#
snarfed
ideally we'll need to account for redirects, ignore scheme, etc, but easy to ignore those for now
#
aaronpk
i hope we don't have to follow redirects
#
aaronpk
i'm going to give this a shot
#
snarfed
great!
#
snarfed
thanks for iterating
#
aaronpk
oy, the h-cite case is hard
snarfed1 and wolftune joined the channel
#
aaronpk
shoot, I don't think the ruby parser is doing this right
#
aaronpk
well the parser part is fine, but all the helper methods aren't
#
aaronpk
i'm gonna have to drop down to the to_hash version of this, darn
#
snarfed1
sounds like the same reason i had to make that maybe_get fn
snarfed joined the channel
#
aaronpk
yeah, the ruby convenience methods are a little finicky
#
aaronpk
i think i've got it!
#
aaronpk
do you have an example URL that uses h-as-like?
#
snarfed
specifically? ie not just u-like-of
#
snarfed
i don't think so
#
aaronpk
hm, it's an additional check in the code
#
aaronpk
"... or entry.format_types.member? 'h-as-like'"
#
aaronpk
i'm tempted to just take that out
danlyke_ joined the channel
#
snarfed
oh i see
#
snarfed
yeah, if we think the h-as-* are sufficiently rare, then sure
#
aaronpk
this is only going to break if there is an h-as-* that does not also include the URL in a like-of property
#
aaronpk
which seems unlikely
#
snarfed
yup, sgtm
#
aaronpk
ok i'm pretty sure this is gonna work!
#
aaronpk
works for my complicated h-cite reply
#
snarfed
bridgy markup fix will be up in a min
#
aaronpk
i need an example of someone who shows reply context but doesn't use h-cite
#
aaronpk
oh i think i can fake it for the test
#
snarfed
deployed mine
#
snarfed
looking fwd to seeing your to_hash approach to the broken dynamic methods (vs the maybe_get)
#
aaronpk
so close
wolftune joined the channel
#
aaronpk
okay deploying it now!
#
snarfed
fingers crossed!
#
snarfed
goes to fave and RT random tweets
#
aaronpk
heh yeah
#
aaronpk
i haven't tested it with bridgy URLs yet
#
snarfed
heh ok
#
aaronpk
only native posts like my replies, or ben's replies
#
aaronpk
so we'll see what hapens
#
aaronpk
hah... just realized twitter actualy generates these types of notifications already
#
aaronpk
"Dora favorited a tweet you were mentioned in"
#
snarfed
aha yes
#
aaronpk
nice!!
#
snarfed
so much nicer
#
snarfed
now bridgy will no longer generate a low background buzz of annoyance here
#
snarfed
thanks again!
danlyke_ joined the channel
#
aaronpk
hooray!
#
@corps_winford
RT @JaimieHollick Sometimes I feel this way, especially during the sprint of #nanowrimo #amwriting #indieauth
(twitter.com/_/status/543647064664834049)
#
KartikPrabhu
Loqi: messages?
npdoty, eschnou, KartikPrabhu, michielbdejong, dns53, snarfed and snarfed1 joined the channel
#
Loqi
[mention] yatil posted 'Heinz Wittenbrink hat ein ”žvorsichtiges Plädoyer für eine Doppelstrategie“ zum Posten von Beiträgen auf Medium und einem eigenen Blog veröff...' linking to https://indiewebcamp.com (http://yatil.de/medium-vs-blog-ein-pladoyer-furs-eigene-blog/)
#
Loqi
[mention] yatil posted 'Heinz Wittenbrink hat ein ”žvorsichtiges Plädoyer für eine Doppelstrategie“ zum Posten von Beiträgen auf Medium und einem eigenen Blog veröff...' linking to https://indiewebcamp.com/POSSE (http://yatil.de/medium-vs-blog-ein-pladoyer-furs-eigene-blog/)
danlyke_ and cmhobbs joined the channel
#
aaronpk
oh you know what... this notification sentence really should be returned via the API as well
#
aaronpk
at which point I might drop native webmention procesing from p3k and use webmention.io's instead
#
aaronpk
because I don't wanna do all the parsing work twice
Pea1, benjamin-agaric and wolftune joined the channel
#
aaronpk
!tell bret I fixed the CORS headers and content type on webmention.io and now your comments are displaying again!
#
Loqi
Ok, I'll tell them that when I see them next
wolftune and reedstrm joined the channel
sub-lunar, memecake, gRegor`, richardlitt, danlyke_, cmhobbs, emmak, tantek, KartikPrabhu, PMurphs and Pea1 joined the channel