2014-05-23 UTC
# 00:02 barnabywalters aaronpk: do you mind me diving into php-comments and splitting the logic up into a bunch of little functions?
# 00:02 barnabywalters a lot of those functions are super handy for reader building, but the whole algorithm isn’t suitable
# 00:03 barnabywalters I’m pretty sure the only thing Taproot’s (unpublished) microformat-flattening code has which this doesn’t is HTML purification
# 00:06 aaronpk if you think it makes sense to rename the package too I'd consider it
jedahan joined the channel
# 00:08 barnabywalters I think what is really needed is a package which, given microformats, gives you a foolproof, purified, normalised, flat list of arrays
# 00:09 aaronpk yeah in my experience there ends up being still too many cases to handle when reading raw microformats, it's very useful to have a library that cleans things up a bit
# 00:09 Loqi aaronpk meant to say: yeah in my experience there ends up being still too many cases to handle when reading raw microformats, it's very useful to have a library that normalizes a bit
KartikPrabhu joined the channel
# 00:09 Loqi barnabywalters meant to say: microformatcleans things up are eacleans things upy to concleans things upume, but not eacleans things upy enough
# 00:10 tantek the "cleaning things up" and "normalizing" tends to be application specific
# 00:10 tantek e.g. being interested in a "foolproof, purified, normalised, flat list of arrays"
# 00:10 aaronpk just that in practice, because there are multiple ways to indicate the author of an h-entry (full h-card, string, etc) you have to do a bunch of stuff
# 00:11 tantek barnabywalters: if you were instead reading microformats from a page to get a hierarchical comment thread - you'd want hierarchy in your structure as in the page
# 00:11 tantek right, with the use-cases you've implemented so far
# 00:11 barnabywalters tantek: by “flattened” I’m referring to properties rather than nested structures
# 00:13 aaronpk so php-comments is a thing that does all the normalization of the raw mf2 data and gives me exactly the properties I want in exactly the place I want
# 00:13 aaronpk and after I wrote that I immediately realized it would be useful for more than just rendering comments :)
# 00:13 tantek the "exactly … I want" - is application specific
# 00:13 tantek which is fine, just consider it when naming such utility functions
# 00:14 aaronpk yep, thats why i'm not even remotely suggesting changing the way we publish mf2 data
# 00:14 tantek barnabywalters: might be worth documenting those
# 00:14 tantek sure - but those go beyond simple microformats
# 00:14 tantek and into variants of pages, page/site structure
# 00:15 barnabywalters tantek: that’s plumbing, I’m talking from the point of view of the consumer of the data
luxagraf joined the channel
tilgovi joined the channel
# 00:32 kylewm is someone going to wiki-fy the emmak-aaronpk-barnaby idea about trustless-micropub? it sounds awesome but i'm not quite following it reading the logs, wondering if i should wait for a summary
# 00:34 barnabywalters kylewm, KartikPrabhu: I would but am going to bed now — if no-one has documented it by the morning I’ll do it then
# 00:34 KartikPrabhu cool! thanks... seems it would be useful as I go through micropubbing!
# 00:35 kylewm good night, barnabywalters thanks again for your help this morning/afternoon
kylewm and caseorganic joined the channel
# 00:53 aaronpk please excuse the in-progress documentation on this
jedahan joined the channel
# 00:59 KartikPrabhu aaronpk: it is a little rough on the edges but I'll get through it and suggest edits as I understand more :)
scor, hidgw, tantek, jasonjohnwells, benwerd, dybskiy, ngoldman_, scor_, tabraldes, jtzl_ and kylewm joined the channel
# 01:58 KartikPrabhu has setup a micropub endpoint. now to do authorization with tokens :)
# 01:59 KartikPrabhu aaronpk: should I have https in place before doing all the micropub stuff?
benwerd joined the channel
# 02:07 kylewm will the wiki still recognize someone after they switch from http -> https?
netweb and tgbrun joined the channel
# 02:11 kylewm aaronpk: that writeup helps a lot. still a little confused on this: why do you need a 'comment' access token at all?
# 02:12 kylewm as in, why not just forward the comment data to your own post UI
# 02:13 aaronpk the micropub POST request has comment text and other junk, and it's going to have to store or encode that data before being able to present the request to you
# 02:14 aaronpk I didn't like the idea of acceping completely unauthenticated requests, since that would end up just wasting server resources either encoding or storing these potential comments
# 02:14 aaronpk so I thought the initial untrusted micropub request could require an access token so that your server could flat out reject bogus requests
fmarier joined the channel
# 02:18 kylewm like you store the comment on your site somewhere and then open the interface to edit it
# 02:20 aaronpk i'm out! feel free to edit that or ask questions and i'll get back to it later!
jjuran and kylewm joined the channel
realz, paulcp, tgbrun and luxagraf joined the channel
# 03:33 GWG KartikPrabhu: Is it worth the money?
# 03:37 GWG KartikPrabhu: What is the advantage for your own site?
# 03:39 KartikPrabhu mainly "But if only the “controversial” stuff is private, then privacy is itself suspicious. Thus, privacy should be on by default" -- Tim Bray
nloadholtes joined the channel
# 03:41 kylewm re: "the domain doesn’t match the security certificate (though the connection is still encrypted)"
# 03:42 kylewm I have wondered about that ... self-signed HTTPS seems like better than nothing, but the browser makes it seem like the end of the world
# 03:42 KartikPrabhu kylewm: yeah unsure how that affects things. But i just geenrated a cert from StartSSL
# 03:43 kylewm I mean I understand why, but it'd be nice if it could say "We can't guarantee this site is who they say they are, but the data to them is encrypted"
# 03:46 rascul "it doesn't really matter who issues your certificates, there are no real assurances behind them any way, and nobody checks"
brianloveswords joined the channel
# 03:54 rascul identification is a joke with ssl unless you pay crap tons of money
# 03:54 KartikPrabhu rascul: In that case we should use it, and hope that smarter folks come up with better stuff
# 03:55 rascul yes, do use ssl and https, don't worry overly much on where the certificate came from
# 03:55 rascul unless it's for a big money making site or something
# 03:55 KartikPrabhu re: indentification, of course encryption is good and identification is hard specially wrt privacy. You need to be a very very trusted institution to do identity verification while respecting privacy
# 03:56 rascul i personally don't trust any of the big certificate authorities
# 03:57 KartikPrabhu <sigh> perfect now my http site does not work and https is not verified yet! nice work
# 03:57 rascul they exist for the sole purpose of ripping people off with over priced and under featured ssl certificates
# 03:57 rascul startssl i use for some things though just so browser doesn't complain
# 03:59 rascul "The first draft of HTTP 2.0 is using SPDY as the working base for its specification draft and editing"
# 03:59 KartikPrabhu oh well i guess my site will be down for a bit unless you trust me ;)
# 04:00 rascul cacert.org may also be interesting for some people
luxagraf joined the channel
jedahan joined the channel
# 04:36 KartikPrabhu kylewm: I changed my mind. Firefox's https warning is exactly what it should be. Though it is annoying for us techminded people, it is "best" that 'normal people' be given a strict warning about it since they might not really understand all the risks
# 04:38 GWG I'm getting two extra Class 1 certificates for my server
# 04:40 GWG I don't want every secure request to be redirected to my site
# 04:41 j12t use a self-signed on, and save some money
# 04:41 Loqi j12t meant to say: use a self-signed one, and save some moneey
benwerd joined the channel
# 04:44 KartikPrabhu might be a little late as a warning but hope you took that into account
# 04:44 GWG KartikPrabhu: Do I plan to revoke?
# 04:44 j12t has anybody played around with the gpg stuff that the freedombox folks have put into apache ssl? I don’t know any details
# 04:45 GWG KartikPrabhu: n9n.us has a StartSSL certificate
# 04:45 GWG I was using it to test SPDY at one point
# 04:45 j12t but it’d be cool to have web of trust instead of hierarchical certs for ssl
# 04:47 GWG I didn't maintain it. n9n.us only runs my TT-RSS installation right now, and a test server for theme development
# 04:47 rascul i have yet to do spdy on anything because i'm lazy
# 04:48 GWG rascul: As did I. They've evolved since I last did it
# 04:48 GWG The module isn't compiled in by default
# 05:01 GWG But one time I updated, forgot, and then didn't fix.
# 05:14 aaronpk I have a bunch of ssl sites on my laptop for development
# 05:14 aaronpk It's nicer to be able to get my machine to not throw cert warnings
# 05:27 aaronpk Sure! As long as you have a cert that won't throw warnings for people
# 05:28 rascul my thoghts were if i was gonna do https, why should i bother still with http
glennjones, whaity, eschnou and indie-visitor joined the channel
# 06:10 rjruss so I got a domain earlier today....now I assume I host it on a server somewhere? any cheap ideas for putting that together? looks like there is a billion options
jsilvestre, cweiske, caseorganic and fmarier joined the channel
# 06:30 jjuran It's really cheap, especially if you have low traffic
# 06:31 rjruss I was just looking at bluehost....I'll check out nearlyfreespeech as well...thank you
Jihaisse joined the channel
# 06:41 KartikPrabhu it also lists people who use diff services so you can them about it here
zaal joined the channel
# 06:43 rjruss KartikPrahbu: hey thanks..yeah I was just reading that a minute ago....just trying to get this put together and then see what I want to do from there
# 06:45 KartikPrabhu some things you might want to look at would be 1.cost 2. technology available that you'd like to use like Wordpress/PHP/Python etc... 3. support direct/through some community forum
petermolnar joined the channel
# 06:53 rjruss yeah cost is a big one..would be awesome to go VPS but yeah..cost is a huge factor! The technology available is something I am unclear on at this point..obviously the more flexible the better..but yeah...and support is one I didn't think of ...yeah thats an important one
nloadholtes, LauraJ and eschnou joined the channel
# 07:34 voxpelli rjruss: all the first VPS:es within an app's cluster is free on http://www.heroku.com/ – only downside is that it will pause itself one hour after it was last accessed, but it will start again the next time someone accesses it
pfefferle, tobiastom, Sebastien-L, chrisroos, tpinto, tgbrun, friedcell, ttepasse, alistair and adactio joined the channel
Sebastien-L, chrisroos, adactio, alistair, jonnybarnes, scor, tgbrun, netweb, aboynejames, pbeaulieu, pfefferle, tobiastom_, glennjones, luxagraf and krendil joined the channel
paulcp, chrissaad, pfefferle, gRegor`, j12t, chloeweil, chloeweil_, ttepasse, jasonjohnwells, luxagraf, brianloveswords and caseorganic joined the channel
pfefferle, jedahan and chrissaad joined the channel
sparverius, pbeaulieu, gRegor`_ and iangreenleaf joined the channel
netweb joined the channel
dariusdunlap joined the channel
j12t, gRegor`, barnabywalters, luxagraf, paulcp and tantek joined the channel
emmak joined the channel
pbeaulieu and brianloveswords joined the channel
# 17:00 aaronpk it's about time for webmention.io to start parsing those bridgy links and making these things look better in IRC
# 17:01 Loqi Ok, I'll tell them that when I see them next
# 17:05 Jeena aaronpk, I use it on my website and it works really well.
# 17:05 Jeena The only thing is that it lacks a bit of documentation on how to use it.
paulcp joined the channel
# 17:13 tantek does the ruby mf2 parser have name beyond "ruby mf2 parser" ?
tpinto and whaity joined the channel
luxagraf joined the channel
chloeweil joined the channel
# 17:29 tantek has been creating Lanyrd and Plancast copies for IndieWebCamp 2014
caseorganic joined the channel
# 17:40 tantek let's see if that garners any more sign-ups :)
friedcell, bnvk and tgbrun joined the channel
# 18:03 tantek chloeweil, know anyone who would enjoy IndieWebCamp? We need to get more people for East! (West is beating us by only 1 sign-up more)
# 18:05 chloeweil tantek I’ll have to ask around, and see which nerds I work with are also freedom fighters :)
gRegor`_ and dariusdunlap joined the channel
# 18:07 tantek chloeweil: especially if they fight for their own freedom!
# 18:07 Garbee Yup. :( I may try to come digitally though.
jsilvestre and squeakytoy joined the channel
pbeaulieu joined the channel
# 18:45 bnvk did y'all know in Doctorow's book Little Brother they create software called Indie Net ?
eschnou joined the channel
# 18:51 bnvk tantek: can't find any official links describing- but "IndieNet" is mentioned multiple times a piece of social software being created by these hacker kids
# 18:52 tallpaul 'id,title,clientCorporation,owner,dateAdded, address(\'state\
# 18:52 bnvk not sure when the book was written / first published- according to Amazon, Jan 2013
# 18:59 bnvk tallpaul: ah cool, I didn't see links to the book source, I'm listening to the audio book :)
pbeaulieu, caseorganic and chloeweil joined the channel
tantek joined the channel
paulcp, caseorganic, tantek and KartikPrabhu joined the channel
# 19:53 tantek tallpaul that's awesome! are his books browsable online?
# 19:54 tantek KevinMarks: you know what would be cool? getting Doctorow to add Fragmentions support to craphound.com and then being able to cite stuff in his books by phrase!
# 19:54 tantek (would make for an amazing demo of Fragmentions too)
# 19:55 GWG tantek: Were you the one who told me about the problem with my SSL site?
# 19:56 aaronpk i feel like I might be better off with a parser that just returns the JSON structure and doesn't do all the fancy attribute accessor stuff
# 19:56 GWG Someone was trying to access the site, and got redirected to another site
# 19:56 GWG I can't remember who in here it was
# 19:56 tantek aaropnk - jlsuttles was last maintaining the ruby uf2 parser
# 19:58 aaronpk i like the way barnabywalters did the PHP one, the parser itself just returns the data structure, and then any fancy normalization/access stuff is handled as a separate library
# 19:59 tantek aaronpk: agreed, that's a good separation of concerns ;)
# 19:59 tantek perhaps try @-mentioning @jlsuttles for info on ruby uf2 parser? (like docs etc.)
# 19:59 tantek she did a bunch of work on it at IWC Hollywood last year
iangreenleaf and scor joined the channel
luxagraf joined the channel
# 20:02 GWG KartikPrabhu: You convinced me to add https support.
# 20:02 GWG KartikPrabhu: I'm not forcing it though
# 20:03 GWG Are you thinking of forcing all connections to https?
# 20:06 GWG I would have to resolve some of the conflicts there
chrissaad and paulcp joined the channel
# 20:22 GWG I turned on 4, but the images were insecure
# 20:23 KartikPrabhu tantek: I think the Level 1 and Level 4 descriptions are not clear specially for beginners
# 20:23 KartikPrabhu if you don't already know words like "ciphers" it doesn't make much sense
dariusdunlap joined the channel
# 20:24 kylewm Level 1 is kinda a double-negative. I couldn't tell if barnaby is an example of "not doing the wrong thing"
# 20:24 GWG tantek: people were talking SSL yesterday, I figured...what the heck
# 20:24 tantek same question to you KartikPrabhu - when did you reach level 3? Today?
# 20:26 GWG tantek: Why? Are we gamiefying Indiemark?
# 20:27 tantek GWG how did KartikPrabhu convince you to get to Level 3 level support?
krendil joined the channel
# 20:27 tantek GWG - yes, IndieMark is definitely a gamification attempt
# 20:27 GWG tantek: He said he was doing it, and I said... "Why not?"
# 20:27 tantek for now we're just tracking IndieWeb Examples per technology
# 20:27 GWG I've been meaning to fix that SNI problem with the server for a while anyway
# 20:27 GWG It was redirecting all SSL requests
dariusdunlap and paulcp joined the channel
# 20:37 tantek GWG, KartikPrabhu what do you think *should* happen if you try to go to a site with https and the site doesn't have a certificate?
# 20:37 GWG tantek: Well, ideally not failure
brianloveswords and dariusdunlap_ joined the channel
# 20:38 tantek like my site currently gives the "This Connection is Untrusted" if you try to go via https - because someone else on my VPS is I think serving a self-signed certificate for "localhost"
snarfed joined the channel
# 20:40 snarfed !tell benwerd you and i might be hosting a few HWCs over the next couple months. the event you did at rockit colabs worked well...want to try there again? i'm happy to join as a member, or donate or whatever, if it'll help.
# 20:40 Loqi Ok, I'll tell them that when I see them next
# 20:41 tantek while you're here, snarfed, what do you think a site without SSL support should do when someone tries to reach it via https?
# 20:41 KartikPrabhu tantek: I think the browser giving a strict warning is a good thing for https without cert
# 20:41 snarfed if you don't support ssl, generally you're not listening on port 443, so clients can't connect
# 20:42 GWG How about... "There is no secure version of this site. Would you like to visit the http version?"
# 20:44 KartikPrabhu which is why I think browsers are doing the right thing. A bit alarmist but is there an alternative?
# 20:45 snarfed not really. and it's not really a problem imho. if you just type a domain into a browser, it defaults to http. if the user explicitly types https, and it can't connect, they're probably clueful enough to understand why
# 20:45 GWG There should be a browser protocol to specify the message, perhaps
dariusdunlap_ joined the channel
# 20:46 kylewm KartikPrabhu: GWG: for both of you I get an HTTPS triangle with ! icon that says "This website does not supply ownership information" when i click on it
# 20:46 kylewm just FYI. I'm not sure if that's expected or not
gRegor`_ joined the channel
# 20:47 KartikPrabhu snarfed: re: spdy, make everything "https" by default so end-users don't have to worry about typing http or https
# 20:47 snarfed kylewm: i'm guessing it's a mixed mode error, not a cert misssing details problem
# 20:47 GWG snarfed: I don't do much with SSL.
# 20:48 GWG If I install SPDY, I might force everything to SSL mode
# 20:48 GWG Who needs to load pictures over SSL usually?
# 20:48 tantek snarfed, well if you think refusing a connection is reasonable behavior I'll note that as an xample
# 20:48 tantek GWG if you don't want your pictures to get sniffed via something like Etherpeg
# 20:49 snarfed tantek: yup. another way to find reasonable behavior is, pick big clueful web properties that don't support ssl, see what they do
# 20:49 GWG I haven't gotten to the level where I am worried about it
# 20:49 snarfed re images and ssl, more importantly, you *have* to serve them all over ssl if the page itself is ssl. otherwise users see the mixed mode error, not the padlock you want them to see
# 20:50 KartikPrabhu yes! I now have to somehow change all the http to https in my posts and css :P
dariusdunlap joined the channel
# 20:51 snarfed GWG: KartikPrabhu: common techniques for this are 1) use relative links (ie without scheme hostname) for images on your owns site, so when a page is served over https, the image will be requested over https
# 20:52 snarfed omitting scheme makes it default to the current page's scheme
# 20:52 GWG I need to fix all of that. For now, I'm leaving it.
# 20:52 KartikPrabhu snarfed: yup I did figure out the "no scheme" so have to edit all files now. fixed it for main fonts this morning
# 20:53 kylewm so https Level 1 is kinda like spelling your name correctly on the SAT :)
# 20:54 tantek kylewm: I'm still trying to figure out a good definition of what level 1 https is
# 20:55 kylewm GWG: KartikPrabhu: confirmed that i get the same error message on https://snarfed.org/2014-04-29_9249 ... which has a link to my profile image which is not ssl. so snarfed was right it's a mixed-mode error, not an issue with the cert/configuration
caseorganic joined the channel
# 20:57 tantek snarfed the reason I ask is because e.g. my site gives a browser warning about a self-signed cert for "localhost" and I'm not sure how to prevent that
# 20:57 aaronpk i don't think you can prevent it on a shared host
# 20:58 snarfed tantek: aaronpk: he may be able to just disable serving https altogether for at least his site, though
# 20:58 aaronpk doubtful... unless he has his own IP address which is unlikely
# 20:58 snarfed doubtful? you can definitely do it on a per domain basis in e.g. apache
# 20:59 snarfed aaronpk: actually, i take it back, you may be right
# 20:59 bear valid self signed certs for local host just means getting the details right for the CN field
# 21:00 bear oh - your talking about locahost on a shared server and not dev localhost stuff
# 21:01 tantek hmm - my service says SSL Service: Enabled but I don't seem to be able to turn it off
# 21:02 tantek so I'm trying to create a self signed cert and see what happens
# 21:02 bear the CN just has to match the host name given
# 21:05 GWG KartikPrabhu: Where is the leaderboard?
# 21:05 tantek is tempted to just serve current content over https with a self-signed cert
erikmaarten joined the channel
# 21:13 tantek does this seem like a legit command line? openssl x509 -in certificate.crt -text -noout
# 21:14 bear you should include -days 1024 also rename your csr to certificate.csr
# 21:14 tantek is playing with things he doesn't fully understand
# 21:15 bear tantek - this is a more normal command: openssl x509 -req -days 1024 -in server.csr -signkey server.key -out server.crt
# 21:15 bear KartikPrabhu - instead of having links in your html include the scheme (https://) you can use //
# 21:15 tantek I'm trying to verify by hand whether the self-signed cert that FF got from my server is the same one that my hosting provider claims I have on my site.
# 21:15 bear and the browser will replace the appropriate scheme
barnabywalters joined the channel
# 21:16 bear yes, then you use openssl x509 -in certificate.crt -text -noout
# 21:16 bear (sorry, thought you were generating one)
# 21:17 tantek knowing that the cert is the same one I created on the server
# 21:17 bear you can also get openssl to connect and validate -- openssl s_client -connect yoursite.com:443
# 21:18 tantek and it looks like it checks out (the public key is the same)
# 21:18 tantek from what I understand of PKI, that's a sufficient check that someone hasn't MiTM their own self-signed-cert in there
luxagraf joined the channel
# 21:24 barnabywalters I think it can be made simpler for the commonest case of commenting on an untrusted site
# 21:24 barnabywalters reducing the process to sign-in (with no access token given out) and form submit to your site
# 21:25 aaronpk that might be ok, but i'm worried about opening up a bad attack vector
# 21:27 tantek ok that worked. certificate exception added permanently :)
# 21:28 aaronpk ok so say you know my micropub endpoint and have authenticated my browser session, you're ready to submit the comment to my micropub endpoint
# 21:29 aaronpk your server makes a completely unauthenticated HTTP request to my micropub endpoint containing the comment I entered
# 21:31 tantek now I have to makes sure my RelMeAuth / OAuth dance takes me back to the https URL
# 21:31 aaronpk my server would respond with a Location header which I would then see and preview the comment and post it to my site
caseorganic joined the channel
# 21:32 aaronpk the problem is that because this request is completely unauthenticated, anyone can start making requests there
# 21:33 aaronpk and my server would have to spend resources to handle those even if they're fake
# 21:33 aaronpk sorta, but I don't have to store anything to handle spammy webmentions
# 21:34 barnabywalters aaronpk: that’s one of the reasons the simplified flow removes the need to store anything
# 21:34 aaronpk with an access token I don't need to store spam requests because I can check for an access token first
# 21:34 barnabywalters aaronpk: yeah, but the simplified flow doesn’t need to store *anything*, or give out access tokens
jgee joined the channel
# 21:42 tantek interesting, it looks like the OAuth 1.0a library I'm using with Twitter automatically knew to ask Twitter to return to my https admin URL!
# 21:42 aaronpk tantek: did you just ssl-enable your site? i'm still getting a warning on it
# 21:43 aaronpk oh I see you got a new cert for tantek.com self-signed
# 21:43 aaronpk so at that point you may as well have gotten a free ssl cert from startssl
# 21:45 GWG tantek: I thought you used BBEdit
# 21:48 GWG Pfefferle has been busy updating the Wordpress stuff.
# 21:52 aaronpk I wonder if it'd be worth recommending using the indiewebcamp CA rather than self-signed certs?
# 21:52 aaronpk cause at least then you can add the root cert to your computer and you won't get the SSL warning for your own site
# 21:52 GWG It's free, and no one has to install anything
# 21:52 aaronpk yeah if you don't mind dealing with their horrible UI
# 21:52 aaronpk but if you don't want to do that, and you're considering a self-signed cert, then why not use the indiewebcamp CA instead?
# 21:53 aaronpk like, only use the indiewebcamp CA if you're considering self-signed
# 21:53 GWG aaronpk: I have to deal with it, but my visitors don't
# 21:53 tantek and with that - I'm off to lunch. Yay for https for my admin!
# 21:56 aaronpk I'm saying that at that point, you're better off using the indiewebcamp CA because then at least you can prevent the browser warning on your own computer for your own admin UI rather than having to click to ignore the warning every time
luxagraf and pfefferle joined the channel
snarfed joined the channel
# 22:13 aaronpk but really @pius liked tantek's post which links to the wiki page
glennjones joined the channel
# 22:21 snarfed to get it more precise, i could require rel-syndication links
# 22:21 snarfed but that would kill 95% of current bridgy users, since very few people add those links
# 22:22 aaronpk wait maybe what i'm thinking only works for likes
# 22:22 aaronpk the only URL that should have the "like" property is the tweet URL
# 22:23 aaronpk and doesn't require any sort of additional markup anywhere
# 22:24 snarfed but it loses the part of bridgy that "bridges" likes on silo posts to likes of non-silo posts
# 22:24 aaronpk but don't you also know the canonical URL of the tweet?
# 22:25 aaronpk either from the citation link at the end or from the new backwards discovery?
caseorganic joined the channel
# 22:28 snarfed yes. are you saying, if i see a PSC or PSL, i should interpret that as the original post link, and not include any other links?
# 22:28 aaronpk and if you don't see a PSC or PSL but discovered the tweet the other direction (by looking for rel-syndication on posts linkde from my home page) then the same
# 22:28 snarfed PSCs are very rare, so that wouldn't hurt most users
# 22:30 snarfed the original hurt i worried about was requiring rel-syndication links, which would cut off most users. i know you didn't propose that, though
# 22:30 snarfed so to summarize, if there's a PSC or rel-syndication link, then exclude any other links
caseorganic joined the channel
# 22:31 aaronpk i think so. but let me come at it the other way as well... looking for an example
# 22:33 snarfed i've knowingly chosen to make bridgy promiscuous, but these are good ways to tighten it without overtightening
# 22:34 snarfed a silo post with a link where the author wasn't in favor of the link?
# 22:36 aaronpk yep that's the situation I was trying to prevent with the description
# 22:36 aaronpk I *think* what you last described would cover this
caseorga_ joined the channel
# 22:44 snarfed doh, github @-mentions don't work if they're inside a link
# 22:45 snarfed so, this is probably the right thing to do, and i should probably do it
# 22:45 snarfed it is a bit sad though, since we lose a little backfeed
# 22:45 aaronpk maybe there's a way to keep the other backfeed but not have it be listed as a "like" or "retweet"
# 22:45 snarfed you're right that likes and reposts like the ones you pointed out are technically incorrect….but they're still kinda nice in spirit
# 22:46 aaronpk in that example, @dotpeople's RT of tantek's tweet counts as a full h-entry post that conains a link to dangillmor's article. it's just not an RT of the article
# 22:47 snarfed so should all non original post links be stripped down to just mentions?
# 22:47 snarfed sounds like yes for like and repost, and it's moot for rsvp. and reply …?
# 22:49 snarfed i'm convinced that all three should be downgraded to just mention, but if you want to find an example, go for it
# 22:49 aaronpk ok let's just go with that for now. i might try to dig up an example later
brianloveswords joined the channel
john3213 and caseorganic joined the channel
# 22:59 kylewm snarfed: what about a PSC or PSL or a link that has the same domain as the tweet's author?
luxagraf joined the channel
# 23:02 kylewm snarfed: actually, can you "recognize" a PSL (as opposed to a regular link) now?
# 23:03 snarfed nah, the PSL part is moot, but your suggestion still works
# 23:05 KartikPrabhu snarfed: kylewm: Does bridgy look only for rel=syndication and not u-syndication?
chrissaad joined the channel
# 23:07 snarfed busy right now, sorry, but kylewm can probably help debug
# 23:08 KartikPrabhu so it detected the /2014 page of course but not my original post which does not have a link (due to char limits)
scor and jasonjohnwells joined the channel
# 23:10 kylewm KartikPrabhu: i think it does not know where to find your unfiltered h-feed
# 23:13 kylewm it looks for h-entries, they don't necessarily have to be in an h-feed
# 23:13 kylewm and it follows to the permalink page, since most people seem to publish u-syndication only on the permalink pages
# 23:13 kylewm yes, it has to be the homepage or linked with rel=feed
grantmacken joined the channel
# 23:17 snarfed thanks kylewm! when you get a chance, would you mind adding that explanation to the bridgy about page?
# 23:18 kylewm snarfed: absolutely, I'll add it right near where you added stuff about rel-syndication recently
jjuran, brianloveswords and tgbrun_ joined the channel
# 23:32 KartikPrabhu just realised that his webmentions might not work with the switch to https!
# 23:42 KartikPrabhu kylewm/snarfed: is there a way to tell bridgy to retry/reparse some tweets?
tpinto joined the channel
# 23:45 kylewm KartikPrabhu: unfortunately once it's done the reverse discovery and not found a backlink, it won't look for that tweet again
# 23:46 KartikPrabhu aah ok. just missed a fav from "@indiewebcamp" but that's cool. no biggie :)
luxagraf joined the channel
# 23:46 kylewm hmm, it's possible that the next time it searches you h-feed for a new tweet, it might pick that one up, and then a retry will work :)
# 23:47 snarfed kylewm: you can also go into the datastore viewer in the app engine admin console and delete entities manually so that it will retry
# 23:48 kylewm snarfed: i don't think i have/want permission to do that?
# 23:48 snarfed kylewm: lol. i think you should be able to. and don't worry, it's hard to do unintentionally, and hard to do anything worse than you intend…but i definitely appreciate the caution
# 23:48 snarfed btw, kylewm and KartikPrabhu, what's the status of mf1 backward compatibility in mf2py? does it work? do i need to enable it somehow?
# 23:49 kylewm snarfed: we're at am impasse on it, with BeautifulSoup+htmllib5 it crashed parsing silencematters.com
# 23:49 KartikPrabhu snarfed: haven't merged kylewm's PR because of one test case where bs+html5lib fails. Using lxml parser seems to work though.
# 23:50 snarfed (kylewm: ah, you're right, you only have r/o access in the admin console right now. i can upgrade that if/when you want)
# 23:50 snarfed ok. consider this a vote for figuring that out then, however you want. i'm itching to use it
# 23:50 kylewm snarfed: that's helpful to know, i'd backgrounded it
# 23:51 snarfed it will help more people use bridgy publish, since lots of blog providers automatically include mf1 but not mf2
# 23:58 kylewm wordpress.com has a bunch of javascript inside the 'entry-content' class