2014-04-15 UTC
KartikPrabhu and gRegor` joined the channel
paulcp and pfenwick joined the channel
# 00:41 acegiak aaronpk: did we end up working out what the upset was between my markup and p3k?
scor, benprew, brianloveswords and _6a68 joined the channel
# 01:59 KartikPrabhu while parsing received mentions should I be looking only for the first "h-entry" in the page?
tantek and kbs joined the channel
# 02:09 kbs dunno if this two-bit idea makes sense
# 02:09 kbs but thought it gave the least worst results in general, when looking for the most important h-* block, to find the first one in bfs order
# 02:10 kbs alternately [to steal an idea from readability] can also weigh a h-* block by the number of child blocks with "interesting" properties
# 02:11 kylewm how about the first one by bfs that actually links to you?
# 02:12 kbs kylewm's idea seems better suited for webmention [I ran into mine when generally analyzing pages for h-* blocks, maybe doesn't make sense everywhere]
# 02:14 kylewm kbs: when you say bfs order, is that pre-order or in-order traversal?
# 02:15 kbs my notion of bfs has a fairly unambiguous traversal, perhaps I'm using the terminology wrong :)
# 02:18 kbs just naming conventions I guess - I often have to look it up for the pedants in my life :)
# 02:29 kbs hm, out of curiosity - how are folks finding "the first"? :-)
# 02:30 gRegor` I use the first one returned by the mf2 parser
snarfed joined the channel
# 02:31 KartikPrabhu kbs: the html will automatically have a tree structure. so the first is basically first by bfs
# 02:31 aaronpk The comments presentation link has a lot more info
# 02:31 kbs KartikPrabhu: actually, that's where I ran into trouble - do the selectors tend to find 'the first' by dfs rather than bfs?
# 02:32 KartikPrabhu kbs: oh sorry yeah 'source-order' would be dfs because mf2 parsers use dfs
# 02:33 KartikPrabhu aaronpk: thanks for that too. I can use it to guide my python version :)
scor, tantek, dybskiy and fmarier joined the channel
# 02:49 KartikPrabhu so can a mention be of more than one type? i.e. "repost-of" and "like-of" ?
# 02:54 gRegor` KartikPrabhu: yes
# 02:55 gRegor` That's what I was starting to work on last HWC
# 02:56 gRegor` What I've decided on so far, for the data structure, is a 'type' field that stores whether it's a mention or a reply
# 02:56 gRegor` Then I have boolean fields for 'is like', 'is repost', and 'is rsvp'
# 02:56 gRegor` It's still in progress though, so could change :)
# 02:57 KartikPrabhu gRegor`: I'm just doing parsing as of now. Storing will have to be looked at :)
# 02:57 gRegor` Ah, well it helped inform my parsing
# 02:59 gRegor` I will only display plaintext
benprew joined the channel
# 03:00 gRegor` I might add in auto-linking later
# 03:01 gRegor` But currently I only permit bold and italics pseudo-code in my blog comments, so I'm not limiting webmentions much by showing them plain text.
# 03:01 tantek you can always use the CASSIS autolinker then
# 03:01 gRegor` I have it on Github if you'd like to poke around, though.
# 03:02 gRegor` I've actually adapted your NewBase60 code from cassis to some work projects. :)
# 03:03 tantek yeah I think the function is called auto_link
# 03:03 gRegor` I use it as part of a checksum on password reset links, to ensure the code is valid and within X days.
dybskiy joined the channel
basal joined the channel
kbs, snarfed, paulcp, catsup and benprew joined the channel
# 04:01 KartikPrabhu bear: what would be a good way to introduce mf2py in ronkyuu? submodule?
catsup joined the channel
benprew, kylewm, dybskiy, Phae, tantek, snarfed, tommorris, pfenwick, v0, LauraJ, eschnou, friedcell, Sebastien-L, tobiastom, melvster, basal, krendil, dybskiy_, dybskiy__, jsilvestre, glennjones, jancborchardt, KevinMarks, adactio, ttepasse, pasevin, benprew_, scor, BjornW, equivalentideas, brianloveswords, benward, JonathanNeal, chloeweil and gRegor` joined the channel
jancborchardt, lmjabreu, hugoroyd, snarfed, dybskiy, tahnok and caseorganic joined the channel
# 15:07 aaronpk KartikPrabhu: that is entirely possible. I may have left that out accidentally
benprew joined the channel
# 15:10 aaronpk snarfed: just got the "bridgy's facebook auth expired" notification! that works well!
# 15:11 snarfed annoying that fb expires them, but at least the renewal works
netweb joined the channel
dybskiy and dybskiy_ joined the channel
paulcp, snarfed, iangreenleaf, tantek, dybskiy and npdoty joined the channel
tilgovi, tantek, benprew, brianloveswords, snarfed and caseorganic joined the channel
jedahan joined the channel
dybskiy joined the channel
# 16:58 tantek this: "it’s a nice way to pick up a team of sixty-plus talented engineers schooled in the ways of Twitter and big data"
brianloveswords, squeakytoy, caseorganic and paulcp joined the channel
# 17:25 tantek aaronpk - are you blocked on meta http-equiv to microformats2 parsing model?
# 17:26 XgF Snarfed: A quick thing, you shouldn't be using tag:twitter.com to identify things
# 17:26 XgF that's for twitter to use for their own purposes...
# 17:27 snarfed it wasn't a great idea to start with, and i've never found a use for using tag IDs as opposed to just URLs
# 17:27 snarfed i'd love to rip them out, but they're pretty embedded now and it's not a high priority. ah well
# 17:29 XgF Yeah. I don't see much use for tag URIs. I do like urn:uuid for just randomly assigned "names" though
cnxtrm joined the channel
# 17:30 snarfed in this case, the silos already have ids that i wanted to reuse. the tag URI part was just to make them unique across silos
benprew, barnabywalters and Ravana joined the channel
# 17:44 cnxtrm idk if anybody can help, but I've completed the registration process for my domain. Since doing this I've decided to redirect 80->443(https). Doing this seems to have borken the indyweb login process
# 17:45 barnabywalters if you want to log in with indieauth using https, you have to manually type the whole URL, including HTTPS
# 17:46 cnxtrm so in terms of setting up a website redirecting this way is not good practice?
# 17:46 tantek but for authentication it's considered insecure
# 17:48 cnxtrm goes off to read the references
snarfed joined the channel
_6a68 joined the channel
# 17:55 barnabywalters tantek: I was considering putting it at the top for that reason, decided against it as problems with Twitter have been much more common
# 17:56 tantek agreed that problems with Twitter are more common. however I think it's useful even to Twitter users to quickly read/learn the bit about https
# 17:56 barnabywalters just looking over /repost — given kylewm + others’ implementations, u-repost-of can be moved out of a “counterproposal” section and the whole page reformatted for consistency with other example pages
# 17:57 tantek you were asking for microformats2 parsers to automatically parse http-equiv for you so you didn't need to in your webmention handling
# 17:58 tantek "blocked" - is it impeding/slowing your progress on one of your top personal itches
# 17:58 aaronpk oh right yes, that would be awesome to add that to the microformats parsers
# 17:58 aaronpk I haven't encountered anything on my immediate list that i'm nblocked on because of that tho
benprew joined the channel
# 18:03 tantek (aaronpk - in case you want to save that URL somewhere with your itches/tasks that depend on it, so you can reping me on it when you get there ;) )
LauraJ, curiousjohn, dybskiy and dybskiy_ joined the channel
# 18:15 waterpigs.co.uk edited /repost (+382) "Tidied, promoted u-repost-of to official w/ markup example, added kylewm in-wild example, my own example of displaying reposts" (
view diff )
KartikPrabhu joined the channel
# 18:19 barnabywalters this is a long shot, but: has anyone ever come across a UI which makes posting content with fallback content easy?
# 18:19 barnabywalters examples: img -> alt, SVG in object -> img -> alt, audio/video -> fallback content
# 18:20 barnabywalters I’ve recently been posting a bunch of SVGs, and other images which I’ve put a bunch of effort into creating fallback content for
# 18:20 barnabywalters also, the text-oriented editing makes it really obvious if a post needs fallback
# 18:22 KartikPrabhu though beware it uses some JS to load the correct image so HTML might look very messy
# 18:25 KartikPrabhu Inkscape, and then I use a SVG optimizer plugin to make them 'simpler'
kbs joined the channel
# 18:28 KartikPrabhu scour is good as it also has good commandline for batch processing andworks as a plugin straight into Inkscape
# 18:29 kbs quite tangential to the subject of posting fallback-ed content
# 18:29 kbs but one of the few times javascript comes in handy for communicating interesting ideas through illustrations is Bret Victor's Tangle - something either of you have played with already?
# 18:30 tantek.com edited /deleted (+169) "/* Brainstorming */ subhead HTML meta http-equiv for status and link to respective microformats2 parsing brainstorming item" (
view diff )
# 18:32 kbs (just falls back to static images
# 18:32 kbs but I guess his thinking was more about communicating ideas in interesting ways, rather than what you were discussing :)
# 18:33 KartikPrabhu kbs: I was looking for such a thing to play around with. Maybe I can use it correctly with fallbacks
# 18:33 kbs exeunts left, pursued by layers of fallbacks
# 18:33 kylewm snarfed: reading back from the logs ... I backfilled a bunch of tweets from Twitter. actually kind of funny, I had deleted ~2,000 of them and stored them in a weird Java-specific database in a fit of privacy anxiety. had a chance to restore them when I got my domain :)
caseorganic joined the channel
caseorganic joined the channel
# 18:35 tantek once again hovering just below ~100 people in the channel!
# 18:37 kylewm !tell barnabywalters thanks for the kind words! I've been trying not to mimick p3k and idno too badly, but ... failing :)
# 18:37 Loqi Ok, I'll tell him that when I see him next
# 18:37 tantek KartikPrabhu: we've been bubbling above 100 for about a couple of weeks now
jedahan joined the channel
# 18:38 tantek kylewm - no apologies, keep re-using what works :)
# 18:38 kylewm snarfed: agree that we should move the github conversation back here/wiki. does it make sense to add a "Proposals" section to the indiewebcamp.com/original-post-discovery ?
# 18:40 tantek kylewm - re: original-post-discovery - are there specific steps in the algorithm that you think could be improved? or a completely different algorithm or?
# 18:40 tantek (trying to avoid re-hashing a thread from elsewhere, hoping you can distill it into a quick summary of conclusions)
# 18:40 kylewm we were discussing ways bridgy could do it without an explicit permalink (but with a link to the parent domain)
# 18:42 kylewm options were basically, add a new query endpoint, find better ways of "hiding" permalinks in each silo (a la See Original on facebook), a new webmention parameter, or have a bridgy-specific mechanism where sites notify bridgy of their relationships before hand
# 18:42 kylewm or bridgy could search the parent site for rel=syndication links, I believe
# 18:43 tantek I think that last one is particularly interesting
# 18:43 tantek especially if bridgy simply signed-up to get PuSH notifications from the indieweb site
# 18:43 tantek and then tracked the rel-syndication links from each original post
barnabywalters joined the channel
# 18:44 Loqi barnabywalters: kylewm left you a message 6 minutes ago: thanks for the kind words! I've been trying not to mimick p3k and idno too badly, but ... failing :)
# 18:45 kylewm ah cool, I actually hadn't given that one a lot of thought
# 18:45 tantek Bridgy as PuSH consumer is the lightest weight (least HTTP), and most standards-using approach of those I think
# 18:46 tantek and that helps encourage indieweb sites to support PuSH + h-feed as well
# 18:47 kylewm does anyone do PuSH + h-feed yet? I know there was discussion of it recently
# 18:48 tantek I think barnabywalters does (as part of his indie reader)
# 18:49 tantek I think he blogged a bunch about it recently too
# 18:49 tantek yes I believe barnabywalters is both publishing and consuming h-feed + PuSH
# 18:50 kylewm does it make sense to capture this discussion in original-post-discovery?
# 18:52 tantek though maybe crosslink from an original-post-discovery Brainstorming subsection as well
_6a68 joined the channel
barnabywalters joined the channel
# 19:14 barnabywalters RE publishing+consuming PuSH 0.4 — yes I am publishing it and have a skeleton consumer working, it’s not yet doing anything with the data other than just storing it though, yet
# 19:14 barnabywalters my current UI challenge is: integrate feed reading tightly into taproot, or make it a separate tool?
# 19:16 tantek well, you could add an entry as a few of the rest of us have, and indicate since when you've supported it
# 19:16 tantek checks to see when he started supporting PuSH
paulcp joined the channel
# 19:18 barnabywalters tantek: kylewm: you both mention PuSH support — are you sending pings for ATOM feed files or for the canonical HTML?
# 19:18 tantek samesies, since 2010. looking for actual start date...
paulcp joined the channel
grantmacken, paulcp and pauloppenheim joined the channel
# 19:36 KevinMarks I thought it was showing me invites, ti was showing suggestions
KartikPrabhu joined the channel
snarfed1 joined the channel
# 19:38 KevinMarks there were more people i knew in the suggestions than the invites...
snarfed1 joined the channel
snarfed, KartikPrabhu, ttepasse, glennjones, scor, melvster, krendil and tilgovi joined the channel
# 20:40 kylewm is there a wiki tag for experimental/proposal-type things?
# 20:53 kylewm snarfed: sure! glad to have found something to sink my teeth into that will affect more than just my site :)
# 20:56 snarfed my biggest fear so far: i want this myself too, and that means i'd have to implement it in wordpress
# 20:57 kylewm that's a good argument for the rel-syndication approach yeah?
# 20:58 snarfed i'd really prefer a solution where i don't have to do any extra work
# 20:59 kylewm haha, I'm planning on trying to implement it whichever way we go
# 20:59 kylewm but yeah, even with that, it would mean more code/complexity/maintenance on the bridgy side
# 21:00 snarfed eh that's fine. i'm just avoiding the wordpress parts
# 21:00 snarfed esp since i posse manually, so rel-syndication links aren't as easy
# 21:03 kylewm I'm thinking instead of PuSH consuming, Bridgy would use the appearance of silo content as the trigger to do the search
# 21:04 kylewm since POSSE might happen sometime after the post is published
# 21:05 snarfed if we also say the query is a new endpoint, the change you already wrote in your branch basically implements it
# 21:06 kylewm oh? are we still talking about searching the h-feed for syndication links?
paulcp joined the channel
# 21:07 snarfed no, i was thinking about the original commit(s) you posted a long time ago
# 21:07 snarfed iirc they discovered the endpoint, then queried it directly
# 21:09 snarfed oh, that commit actually returns the query url instead of fetching it. still works though.
# 21:10 kylewm i'm not sure why .. but the query url actually resolves to the final permalink somewhere down the road
# 21:10 kylewm so it showed up in bridgy's UI as the proper permalink
# 21:10 snarfed i aggressively follow redirects and cache the results
# 21:10 kylewm except for 404 cases, where it showed up as the query url
# 21:11 kylewm I think tantek convinced me to try to stick with the existing standards though, unless we get to a point where we absolutely can't
# 21:11 snarfed that's only for sites that advertise a query endpoint, right?
# 21:12 snarfed yeah fair point. an alternative would be for webmention handlers to support silo post targets and lookup internally
# 21:12 kylewm like target=twitter-post&source=twitter-reply ?
# 21:13 kylewm oh except it would be source=bridgy-proxy-of-twitter-reply
# 21:14 snarfed but no one supports the query endpoint either, so nm
# 21:14 kylewm dang, I feel like I went from 0 good options to 3 good options
# 21:15 kylewm abusing webmention that way is a little less explicit than the secondary lookup, but it'd definitely be less work all around
# 21:18 kylewm it's not totally unlike receiving a mention to a shortlink
# 21:20 snarfed it'd be nice to have some discovery component, so that bridgy doesn't constantly send doomed WMs to sites that don't support this
# 21:23 kylewm yeah you're right, that would be a little spammy
# 21:25 snarfed maybe it'd be fine, since we'd have to special case this a bit anyway
# 21:26 snarfed hmm. i'm not sure the wms would go to the root domain. wouldn't source be the bridgy url and target be the silo url?
LauraJ, paulcp and tantek joined the channel
# 21:52 kylewm snarfed: ah yeah, but you'd discover the wm endpoint from the root domain, rather than from the permalink
# 21:52 snarfed ah i see, sure. discovery from root domain, webmention to silo url
# 21:54 snarfed btw, now that we have that wiki page, i'm going to close the bridgy issue. holler if you disagree
paulcp joined the channel
# 22:09 snarfed hey, yeah, why aren't there more indieweb mobile apps?! :P
_6a681, pasevin, KartikPrabhu, kbs and lukebrooker joined the channel
# 22:24 kbs By my definition of 'indieweb' anyway :) think I have a few [though not public] mobile apps
# 22:25 kbs sees some tension between html-as-display and html-as-data as far as robustness of parsing and ease of display goes, between a mobile app and a browser
# 22:25 snarfed yeah, i'm actually open to it too. i think the philosophy is indie first, web second
# 22:29 kbs that's an interesting way of lookng at it
# 22:29 kbs I was myself framing it as indieweb first, html second :)
# 22:31 kbs ie, http still seems quite useful
# 22:31 kbs though it really isn't Ht-tp anymore I guess
j12t joined the channel
fmarier and benprew joined the channel
cnxtrm joined the channel
# 23:02 aaronpk I'm gonna have to make a basic ios twitter clone app aren't i
# 23:02 aaronpk tiny feed reader and micropub client for creating note posts
# 23:03 aaronpk there's probably an open source app.net client I can fork
friedcell joined the channel
_6a68 joined the channel
# 23:56 tantek j12t, do the developers of Mediagoblin selfdogfood it on their own domain?
# 23:59 tantek aaronpk, snarfed, I don't understand - isn't your website its own mobile client?