2014-03-24 UTC
# 00:05 kylewm KartikPrabhu: Oh great! It does seem like it should not be possible to visit things more than once with recursive set to false
# 00:05 KartikPrabhu yes. But I'd like to make sure that is the case before turning off parsed
# 00:08 kylewm KartikPrabhu: also on the subject of BS4, have you seen a problem with classes on the parent html tag? e.g., <html class="a b c d">
# 00:08 kylewm when I do get("class"), it returns 'a b c d' rather than [
# 00:14 kylewm weird, me too! can you try running exactly the same code i am? from bs4 import BeautifulSoup;soup = BeautifulSoup("""<html class="a b c d"></html>""");soup.find("html").get("class")
jonnybarnes and fmarier joined the channel
# 00:35 Loqi Ok, I'll tell them that when I see them next
# 00:36 kylewm KartikPrabhu: well with html5lib installed, it gives me 'a b c d', without it, it gives me a list
# 00:39 KartikPrabhu kylewm: I tried with all 3 parsers here is the result: lxml - ['a', 'b', 'c', 'd']
# 00:40 KartikPrabhu I have not set any default parser in mf2py, so BS picks the best one as lxml > html5lib > html.parser . I am not sure which one is on my server
# 00:41 kylewm I *think* this means I get to file an issue with BeautifulSoup
# 00:44 kylewm good point, I use 3.3 to try to avoid those issues :)
# 00:45 KartikPrabhu yeah I sould migrate my sytem to py 3.3 but I'm not sure how the other parts of my setup will hold
# 00:47 KartikPrabhu kylewm: fyi - I just checked an my server uses the inbuilt python parser
dangillmor and gRegor` joined the channel
# 00:54 kylewm KartikPrabhu: I think when I added BeautifulSoup4, I should have removed html5lib from requirements.txt
# 00:55 KartikPrabhu maybe. BS4 just uses whichever parser it can find in preference order. It is really weird that html5lib has that strange bug
# 00:56 KevinMarks I'm wary that the other parsers won't be as correct for html in the wild, if they are assuming valid xml
# 00:57 kylewm KevinMarks: other parsers won't be as correct as html5lib?
# 00:57 kylewm afaict html5lib does not process attributes into multi-valued attributes at all, i think that is a bs4 thing
# 00:58 KevinMarks I think lxml will not do what browsers do with unclosed elements etc.
# 01:08 KartikPrabhu yup. bs4 is supposed to return a list if more that one class value is found. But using html5lib somehow it returns just a space-separated string for the html tag
# 01:14 KevinMarks Probably the HTML tag is a special case in HTML5lib, unlike the other xml parsers
# 01:14 kylewm KartikPrabhu: KevinMarks: yes definitely it is a special case, if <html> is not provided, it adds it around the other tags
# 01:14 kylewm so <body> is not the top-level tag from bs4's point of view
# 01:17 KartikPrabhu kylewm: actually I think html5lib will also auto-add <body> if it is missing. So I am convinced that html5lib is fine and bs4 is doing somethign weird
# 01:18 kylewm yes, you're right, it adds html, head, and body if they do not exist, and I agree wtih you
# 01:18 kylewm I believe snarfed also uses h-entry on the html tag
# 01:18 KartikPrabhu so hopefully this gets fixed, and I'd be happy to specify html5lib as the parser for mf2py
# 01:19 kylewm thanks again for your help, off to make dinner
# 01:44 KartikPrabhu dangillmor: you mean that "likes" should not show as actual comments?
ttepasse joined the channel
# 02:00 dangillmor KartikPrabhu: i like that. wonder if I can get this to work in the WordPress plugin...
# 02:01 KartikPrabhu it seems that right now everything on your site is being saved a comment, you might need to hack around the template/plugin
ttepasse and snarfed joined the channel
# 02:10 Loqi snarfed: KartikPrabhu left you a message 1 hour, 35 minutes ago: how much does bridgy/publish truncate for tweets?
# 02:12 dangillmor have to take a look, see if i can understand how it's working.
ttepasse and dariusdunlap joined the channel
# 02:26 snarfed KartikPrabhu: good question! yeah, i saw your preview that was missing the link. i might be counting chars in links before they get shortened
# 02:32 snarfed oh, no, should be just once, but it's long enough
iangreenleaf, ttepasse, brianloveswords and KevinMarks joined the channel
ttepasse, snarfed, dariusdunlap and krendil joined the channel
ttepasse joined the channel
ttepasse joined the channel
ttepasse joined the channel
ttepasse, ddysart and basal joined the channel
melvster joined the channel
ttepasse joined the channel
# 05:18 aaronpk !tell snarfed did you say there was a way to get Bridgy to generate the HTML for a silo post given a URL even if Bridgy didn't discover it itself? I'm thinking about writing a script to bulk import mentions of my Instagram photos that aren't discoverable by Bridgy
# 05:19 Loqi Ok, I'll tell them that when I see them next
dariusdunlap joined the channel
npdoty joined the channel
bnvk_, glennjones_, KevinMarks, eschnou, KevinMarks_, Sebastien-L, ttepasse, dvirsky, pfenwick, cweiske, Garbee, pasevin, jjuran_ and moofy joined the channel
ubax, pfenwick1, glennjones, LauraJ and voxpelli joined the channel
# 10:37 cweiske does someone have a GUI example for displaying linkback statuses for a website?
# 10:38 voxpelli Just out of curiosity – anyone here following what is happening at the w3cpayment conference or doing any work in that area?
jonnybarnes, bnvk, pasevin, voxpelli, LauraJ, melvster and glennjones joined the channel
glennjones, chloeweil, tobiastom, KevinMarks and tahnok joined the channel
# 14:09 KevinMarks Voxpelli not seen that, no. There is some payment discussion here
glennjones, _6a68, Sebastien-L, bnvk, scor, Guest91157, cjuehring and snarfed joined the channel
# 14:37 Loqi snarfed: aaronpk left you a message 9 hours, 18 minutes ago: did you say there was a way to get Bridgy to generate the HTML for a silo post given a URL even if Bridgy didn't discover it itself? I'm thinking about writing a script to bulk import mentions of my Instagram photos that aren't discoverable by Bridgy
# 14:39 snarfed aaronpk: sure! dirty little secret: bridgy always uses silo apis when it serves silo posts as mf2, ie it never uses its own data (right now at least)
# 14:39 snarfed so you can replace the post/comment ids in those urls with others
pfenwick, eschnou, moofy, gRegor`_ and ozten joined the channel
# 15:24 cweiske all people that send out pingbacks/webmentions: do you visualize the results somehow? show errors?
# 15:28 kylewm cweiske: I send back success/failure + a text description and show it in the UI while posting is happening
scor joined the channel
# 15:31 cweiske kylewm, do you re-try pinging when the server responded with e.g. a "500 Internal Server error"?
# 15:39 cweiske you already sent the webmention, so stapibas probably sent a "already registered"
# 15:41 cweiske 'Linkback from ' . $source . ' to ' . $target . ' has already been registered.',
# 15:42 kylewm oh, cool! I need to print out more detailed info
bnvk_ joined the channel
# 15:45 cweiske apart from that, my linkback server tries 5 times if a pingback was unsuccessful because of technical errors
# 15:45 cweiske unavailablity of remote site or status code 500 on the remote side
# 15:47 kylewm Printing out more detailed info, confirming that I get "Failure for http://cweiske.de/: Status code: 400, Response: {"error":"already_registered","error_description":"Linkback from http:\/\/kylewm.com\/reply\/2014\/03\/24\/2 to http:\/\/cweiske.de\/ has already been registered."}
"
# 15:47 kylewm cweiske: does it try 5 times right away, or delayed somehow?
# 15:48 kylewm so that is an area where I am still pretty lost...haven't figured out how to put things on a queue/schedule
# 15:48 kylewm right now everything on my server happens in direct response to a request
# 15:49 cweiske when I post, I notify my pingback server that I want him to ping
# 15:49 cweiske the PB server currently checks every 5 minutes if there is something to ping IIRC
# 15:50 kylewm I would like to do something like that with everything after making the initial post...syndicating and all that
KevinMarks joined the channel
# 15:50 cweiske right now I'm building an interface to check the linkback status of all links on a page
# 16:02 kylewm cweiske: thanks that is useful! 'code' is not a HTTP status though?
# 16:05 kylewm hopefully other people will show off their interfaces too
pasevin and barnabywalters joined the channel
# 16:12 Loqi barnabywalters: bret left you a message on 3/23 at 4:25pm: thats super rad.
# 16:12 Loqi barnabywalters: bnvk left you a message on 3/23 at 4:33pm: that notification flow is awesomesauce!
KevinMarks, eschnou, ben_thatmustbeme, tilgovi, CheckDavid and scor joined the channel
# 16:59 cweiske barnabywalters, do you visualize linkback responses in taproot?
glennjones joined the channel
snarfed joined the channel
benprew joined the channel
# 17:15 barnabywalters there are a bunch of great plugins which will get your blog sending and receiving cross-site comments in no time
# 17:22 ben_thatmustbeme barnabywalters: thanks, I hadn't yet decided if I was going to keep wordpress or not since I want to do some hacking on things
# 17:22 snarfed i also considered a "resend webmentio now" button for failed webmentions, but i don't know how many people would care. holler if you want that.
caseorganic joined the channel
# 17:24 kylewm snarfed, I've maybe twice gone back and resent one manually with curl
# 17:24 kylewm it would be really nice to have it in the UI but not critical
# 17:24 snarfed cweiske: oh i see. yeah, bridgy already retries for 3 days, backing off to once/hr
# 17:25 snarfed the "retry now" button would just be an additional UI feature, like "poll now," if you want to retry *right now*
benwerd joined the channel
# 17:26 kylewm I might back that off a little to "nice" instead of "really nice" :)
# 17:26 ben_thatmustbeme huh, IndieAuth Links plugin for wordpress 0.2 ... The package could not be installed. No valid plugins were found.
# 17:27 snarfed ben_thatmustbeme: i wouldn't worry about that plugin. just add rel-me links and you're good.
# 17:27 ben_thatmustbeme yeah, for me i'm not too worried, would be nice to fix it for others though I suppose
# 17:27 snarfed kylewm: the bridgy wms you resent with curl, do you remember how long afterward that was?
tantek, iangreenleaf, LauraJ and scor joined the channel
# 17:41 ben_thatmustbeme I really should be working I suppose, rather then messing with my personal site... not much to do at work today though
# 17:47 tantek benwerd, hober, iboxifoo, KevinMarks, Phae, snarfed ^^^
# 17:48 tantek aaronpk, caseorganic ^^^ - hoping you can sync-up with dietrich & skinny also
# 17:51 tantek aaronpk - the new mozpdx setup worked really well with the vidyo
# 17:52 aaronpk ok. but I have a feeling it'll work just as well from here with the camera I used at indiewebcampsf
# 17:53 tantek aaronpk - you have to see how awesome skinny had it setup
# 17:54 tantek anyway - RSVP for now until we can get ahold of skinny and/or dietrich to figure out details
KartikPrabhu joined the channel
# 17:55 benwerd tantek (& others) - might you be up for some food / drink afterwards?
# 17:55 tantek they're not that much interesting to watch per se - since they're mostly in person discussions
# 17:56 aaronpk we haven't recorded video at homebrew before... but there are copious notes usually
# 17:56 benwerd cool. it's been a while, would be nice to catch up :)
# 17:56 tantek we might share a live viewing URL to see if that's interesting
# 17:56 tantek usually Kevinmarks does an amazing job of live-tweeting
# 17:59 ben_thatmustbeme I hope so. I really got interested in the project from hearing KevinMarks talk about your discussions on managing contacts on TWIG. I've actually been thinking of that problem for some time
bnvk joined the channel
# 18:03 tantek ben_thatmustbeme: oh interesting - well I hope to have another blog post in that series published soon (like today with any luck)
# 18:07 tantek benwerd - want to post an Indie Event (and POSSE FB copy) for this week's HWC meetup?
# 18:10 tantek bnvk - have you seen all the Mozilla work on open app stores?
# 18:11 tantek lots there you could document and re-use, likely as-is, since it is all web-based (rather than proprietary silo based)
# 18:12 bnvk tantek: nah, I haven't seen it, any links you can point me to?
# 18:12 tantek bnvk - it's only shipping on every FirefoxOS phone worldwide ;)
# 18:13 bnvk Yes. I have not seen anyone who has one IRL over here in Europe
# 18:14 tantek anyway, it's deployed, it's shipping, it's based on open web, so take that however you will
# 18:16 bnvk yah, only thing is it's fundamentally quite different than what my app store idea is trying to do- mainly have support for not only HTML5 web apps
# 18:16 tantek bnvk - really? what's the point of that in the scope of indiewebcamp.com/store?
# 18:17 bnvk or rather, software, that is connected to the internet
# 18:17 tantek so you've come around to the client vs. server distinction I was trying to convince you of :)
# 18:18 bnvk I still like the idea of at least trying to view non cloud servers as viable hosting ;)
# 18:18 tantek your point: " it's fundamentally quite different than what my app store idea is trying to do" is in contrast with the examples of apple/android app store that you listed top of the page
# 18:18 tantek my point is that the open web app store is like those except based on open web tech, so is a better example from the perspective of figuring something out for the server
# 18:18 bnvk right, installing Wordpress or IDNO on hardware a user controls, e.g. a Pi, or a VPS
# 18:19 tantek which seems like it is client side focused given that you open with the apple/android examples
# 18:19 tantek I was waiting for you to curate accordingly :)
# 18:20 tantek (since you were/are making so much progress on the page, and clearly had a specific vision for it)
# 18:20 bnvk hehe, yes definitely been putting in a bit of work as of late on it
# 18:21 tantek I'm definitely excited by the prospect of server-side installing/setup being made "as easy" as modern clientside app stores.
# 18:21 tantek it's definitely one of those "hard problems" that will make the indieweb easier for everyone (gen 1,2, etc.)
# 18:22 bnvk I'll add a block right now to try to better explain
KevinMarks joined the channel
ttepasse joined the channel
# 18:42 kylewm I think I'm finally going to have to thrown in the towel on auto-generating text for tweets ... notes with several links etc
# 18:43 KartikPrabhu kylewm: why not auto-generate but keep ability to edit the auto-gen one. That way it'll work mostly and special cases can be handled separately
# 18:45 kylewm KartikPrabhu: even better, but I will have to translate the shortener to JS!
# 18:47 benwerd tantek - was at UPS store (ugh), will post event now
# 18:49 dietrich hm, google calendar cannot import the events ics linked from v
# 18:51 tantek dietrich - if you can find a unique useragent for Google Calendar I'll see what I can do about updating h2vx robots
# 18:52 tantek as it was, Google's spiders were being too stupid to not just keep trying to crawl ICS feeds from every h2vx link on there (when they should just be indexing the hCalendar and h-calendar directly)
# 18:53 tantek which would request the ICS and then just dump it on the floor (since Google Web search does nothing with ICS either)
# 18:54 tantek right, this is a Google deficiency on multiple fronts as noted
# 18:55 tantek Google is not very traffic/CPU etc. friendly to small services
# 18:55 tantek unintentionally I'm sure, but neglectful at best
fmarier and ttepasse joined the channel
caseorganic joined the channel
ttepasse joined the channel
# 19:23 snarfed not that either of those will solve this problem, just related
tantek joined the channel
# 19:25 KartikPrabhu snarfed: yeah I have seen those. but this seems to be a small h-card vs detailed h-card problem
# 19:26 snarfed KartikPrabhu: right. i wonder which one the representative h-card alg would return
nagaway joined the channel
# 19:27 kylewm snarfed: I don't think I had, and I do think that algorithm answers my question. use the first h-card with u-url == rel=me, or something along those lines
# 19:28 tantek tommorris could choose to combine them by giving them the same u-uid
# 19:29 KartikPrabhu kylewm, snarfed, tantek: yes the representative algorithm picks the more detailed one! very nice
# 19:30 tantek Then we just have to document application-level merging of h-* objects with the same u-uid
# 19:32 tantek I think barnabywalters has a representative hcard function
tahnok, dvirsky, gRegor`_, KevinMarks and eschnou joined the channel
# 19:47 gRegor`_ Is Nickserv down?
# 19:49 gRegor`_ Hm. I have a registered nick, but someone else is using it somehow.
# 19:50 KartikPrabhu gRegor`_: you can just do "/nick yourname" to change it for this channel at least
# 19:51 gRegor`_ I can't. Someone else is using it.
# 19:51 gRegor`_ That's what is odd.
# 19:51 aaronpk are you sure you're not connected from somewhere else?
# 19:51 gRegor`_ I thought if you /nick to a registered nick, it booted you after a few minutes if you didn't ident
# 19:52 gRegor`_ Yeah, it's definitely not me. :) /whois gRegor`
# 19:53 gRegor`_ This all seems correct, too:
# 19:53 gRegor`_ NickServ Information on gRegor` (account gRegorLove):
# 19:53 gRegor`_ NickServ Registered : Feb 07 21:53:22 2014 (6 weeks, 2 days, 21:57:42 ago)
# 19:53 gRegor`_ NickServ User reg. : Feb 03 19:13:22 2014 (7 weeks, 0 days, 00:37:42 ago)
# 19:53 gRegor`_ NickServ Last addr : ~me@71.239.44.58
# 19:53 gRegor`_ NickServ Last seen : Mar 24 01:22:57 2014 (18 hours, 28 minutes, 7 seconds ago)
# 19:53 gRegor`_ NickServ Flags : HideMail
# 19:53 gRegor`_ NickServ *** End of Info ***
# 19:54 gRegor`_ Guess I can use my canonical nick
# 19:55 gRegorLove Nickserv just told me there was one failed login attempt by that other Gregor since I logged in last.
# 20:06 kylewm KartikPrabhu: t'would seem mf2py doesn't get <a rel='me'> links, only <link rel='me'> links?
KevinMarks joined the channel
# 20:10 tommorris kylewm: I’ll have a look at mf2py real soon now. crazy busy
# 20:13 KartikPrabhu kylewm: that is interesting. I suspect it has to do with the whole "parsed" set thing. I'll open an issue
# 20:13 kylewm tommorris: thanks! my intent is not to pile work on anyone else, just to catalog issues if I come across them that one of us can fix some day :)
# 20:15 tommorris cool. I’m gonna merge BS in soon too. again, need some time in my life. ;)
# 20:15 gRegor` Figured it out. You can issue a "release" command to NickServ to force the nick to be released.
# 20:22 gRegor` Nice domain, ben_thatmustbeme. And welcome aboard. :)
yaf joined the channel
_6a68 joined the channel
marcthiele, tantek, caseorganic and snarfed joined the channel
basal joined the channel
# 20:51 aaronpk ben_thatmustbeme: no it's an ssl issue right now. when I moved indieauth to the new indiewebcamp server it'll be fixed
gRegor`, fmarier, pasevin, chloeweil, kbs, bnvk, KevinMarks_ and pauloppenheim joined the channel
lukebrooker, ttepasse, tantek and basal joined the channel
ttepasse, pasevin, caseorganic, tilgovi and scor joined the channel
KartikPrabhu and benprew joined the channel
# 23:29 Loqi pauloppenheim meant to say: excitement
# 23:30 pauloppenheim when the AI apocalypse rains down upon us, i will at least know that I am welcomed by our new robot overlords.
# 23:32 gRegor` considers a counter-criticism section for /feed
# 23:33 gRegor` On my site I've chosen a single-post view for articles. When I add notes, I will display several on a page, but I don't intend to change the article's display.
# 23:34 gRegor` Seems rather intensive for an h-feed reader to need to hit, say, 10 pages to get my most recent 10 articles.
# 23:34 aaronpk I don't think that's a counter-criticism of RSS and Atom, that's more of a criticism of h-feed
# 23:35 aaronpk also that isn't really that intensive, cause imagine someone visiting your site, they'd have to click 10 times to see all 10 articles anyway
# 23:35 gRegor` Vs fetching one RSS feed, though, is what I meant.
# 23:36 snarfed one approach would be to include full article contents in the h-feed and hide parts via display: none
# 23:36 aaronpk otoh you have people like tantek who had to reduce the number of posts in his atom feed because it was taking up too much bandwidth having the file downloaded in its entirety each time
# 23:36 aaronpk so perhaps the 10 pageloads is actually a feature
# 23:37 snarfed eh. i think we've overoptimized solving for that specific problem
# 23:37 aaronpk i'm just saying i wouldn't worry too much about that either way until you actually have a problem with it one way or the other
pfenwick joined the channel
# 23:38 aaronpk tantek had a real issue, so had to address it. i doubt you will have an issue with separate pageloads or with too high bandwidth usage on a single large file for a while, so best to spend your energy elsewhere :)
# 23:38 snarfed sure. gRegor` 's is actually reasonable though. if he doesn't render full article contents in his h-feed reader, and you subscribe to him in your selfoss reader, you wouldn't get full contents either
# 23:38 aaronpk as a reader, that's fine with me, i'll click the thing if the headline is interesting
# 23:39 snarfed right, but that's you. i'd rather have the full contents.
# 23:39 gRegor` I'm not concerned about bandwidth currently. Mostly just approaching it from "RSS vs h-feed". You're right, it is more a criticism of h-feed.
# 23:39 aaronpk since the preference is really on the consumer, it doesn't really seem like something the publisher should decide
# 23:39 gRegor` aaronpk: Your h-feed reader only pulls in titles currently?
pasevin joined the channel
# 23:40 aaronpk no, but some people only publish titles so I only see the titles for theirs
# 23:40 snarfed (re tantek's issue, i haven't researched it thoroughly, but i have to admit i'm a bit skeptical. most feed readers are pretty good with head requests and etags these days. i wonder how easily he could have cut down that bandwidth problem with some head handling and caching headers)
# 23:40 aaronpk and I don't really mind it, in fact I prefer a title+picture or title+summary over seeing the full article in my reader anyway
krendil joined the channel
# 23:56 tantek it's more work to do an RSS feed than an h-feed - why are you making more work for yourself?
# 23:56 tantek if you don't want to put a feed on your home page - no problem, then you can make it a separate URL
# 23:57 tantek there is still no excuse for a) separate format (i.e. just use HTML+h-feed), b) inefficiencies of all the XML based formats
# 23:57 tantek whether home page or separate URL for your "feed" is semi-orthogonal to the format issue
# 23:58 tantek except that XML based formats FORCE you to have a separate URL
# 23:58 tantek whereas if you just use h-feed, it's your CHOICE whether to put your feed on your home page or on a separate URL
# 23:59 tantek gregor - I will put forth that there is zero advantage besides legacy consumer support for RSS/Atom/XML/RDF (which is a non-trivial advantage, but that's it)
# 23:59 tantek for all other reasons, "just" using h-feed, whether on your home page or on a separate feed.html page, is much easier, more reliable etc.