#tantekhah re: Flickr selling CC-BY photos. And this is why I set my Flickr photos to CC-BY-NC back in the day (2005) - had a feeling something like this might eventually happen.
#tantekKartikPrabhu: re: rel-bookmark testing - I believe I have a bunch of old blog posts in hAtom with rel-bookmark etc. - do you have a place where I can submit URLs and see the backcompat parsed uf2 JSON response?
#Loqithedod: tantek_____ left you a message 1 day ago: are you now posting all notes/tweets/replies from your own site dubiousdod.org/indie and never directly to Twitter ? if so, add yourself to /ownyourdata#IndieWeb_Examples !
j12t, caseorganic, tantek, thedod_, snarfed and chrissaad joined the channel
#owen1tantek: i noticed you posted some stuff, but my pubsub client didn't recieve any messege ):
#owen1also something that i noticed - when i read your website, it's not easy to understand the conversations. let's take as an example the last item there.
#owen1@sil reasonable request. I wrote @cassisjs ellipsize_to_word for that.
#@silCan I make text-overflow: ellipsis not break in the middle of a word? So "too long" becomes "too..." even if "too lo..." would fit? (twitter.com/_/status/537627213789949952)
#Loqitantek: KartikPrabhu left you a message 35 minutes ago: do I own my data on mobile if I don't use twitter on my mobile and don't post on my website from mobile either?
#GWGI was mentioning I'm moving my site to a new VPS. And I was thinking of writing about it. But someone pointed out that writing about how you implement your server configuration was a potential security issue. I don't think it is, to a point, but how does one draw that line
#tantekGWG it is a potential security issue because every piece of information you give a potential attacker about your setup helps narrow the space of weaknesses to explore.
#GWGtantek: So, how do you share information without endangering yourself in that regard?
#tantekyou decide you're not a worthy target, or you decide you'd rather share publicly in the hopes that friends will warn you about any flaws before an attacker exploits them
#Loqisecurity in the context of the indieweb may refer to security concerns regarding personal domains, web hosting, https setup, private data, identity etc http://indiewebcamp.com/security
#GWGtantek: There is also the third part...never post 100% of information.
#GWGSuch as, do not post complete configuration files, only excerpts.
yakker joined the channel
#tantek.comedited /security (+226) "emojicon, Web Hosting, fix headings, remove "content injection" handwavy theoretical. security should focus first on reproducible problems, rather than be distracted by hypotheticals" (view diff)
#Loqisecurity in the context of the indieweb may refer to security concerns regarding personal domains, web hosting, https setup, private data, identity etc http://indiewebcamp.com/%F0%9F%94%92
#tantekowen1 - right - you need your own personal domain to login
#owen1i click login, it takes me to indieautho.com, i sign the challenge with my GPG key, i am sent to indieauth.com/success, it tells me 'You Successfully Authenticated!' i navigate to http://indiewebcamp.com/irc-people but i am not logged-in.
#owen1i would think that the redirect should be to the last url where i left indiecamp, which in my case /irc-people
#aquariusCan I delegate webmentions to a different URL? That is: declare "if you're planning to send a WM to url A, instead please send them with a target of url B"?
#aquariusI think I can't -- nothing in the spec about it
#aquariusobviously I can build what amounts to a "proxy endpoint" where A says its endpoint is the proxy and the proxy connects to B's endpoint and passes the source and a target of B.
#kylewmaquarius: what's the use-case for redirecting wms from one target to another? i know it has come up with http vs. https before, but that generally feels like something that should be handled by the endpoint
#Loqikylewm: KartikPrabhu left you a message on 11/29 at 3:29pm: just confirming that the rel-bookmark change was field tested before I merge the changes
#aquariuskylewm, use case: I'd quite like WMs sent to hash-for-vouch to actually end up on my blog post *describing* hash-for-vouch, because the blog is all set up to deal with WMs, display them, etc.
#snarfedaquarius: kylewm: also, this would be mainly for when you don't own the wm endpoint, right? since if you do, you can just redirect to target B yourself
#aquariussnarfed, yep, although I think that hardcoding magic handling for certain URLs into the endpoint is Not The Right Way To Do It :)
#kylewmaquarius: ah! that makes sense. and you're right that there's no general way to do that if you don't control the endpoint
#aquariusbut my proxy idea won't work anyway because the eventual endpoint will check the source and say "you do not link to the target"
#aquariusso I think you're right in that it'd need special handling
#kylewmCould your blog post query webmention.herokuapp.com for the URL of the vouch-for-hash page when collecting comments?
#aquariusspecifically, an extra <link> element declaring what the "replacement" target is for this URL, every wm sender to send the "original" target and the "replaced" target, and the endpoint to understand this. This is all a lot of work for a pretty edgy edge case ;)
#aquariuscan't guarantee my activity will stay at this level, but I thought yesterday's discussion was interesting and worthwhile putting together a little service to do what we were talking about
#aquariusI think I'm less flattered by the sites which just steal the stuff from my rss feed and display it on a page with ads around it, although I gave up the whack-a-mole game of trying to *stop* this sort of thing years ago :)
#aaronpkaquarius: wow awesome that you launched hash-for-vouch already!
#aaronpkI'm trying it out right now. it's doing a aloooooot of work!
#aquariusaaronpk, spare day, and it's basically a one liner to verify a hash, and everything else is just turning it into a useful service ;)
#aaronpkheh yeah. I often find it's the "everything else is just turning it into a service" that takes a lot of time :)
#aaronpkI'm doing something simliar right now actually
#aaronpkwebdav -> flickr, then next is webdav -> micropub
cweiske joined the channel
#aquariusha, yeah. Work in the browser takes a bit longer, because it's a slightly slower environment, and because I'm paranoid about not locking up the UI so I chunk the work up quite a bit.
#aquariusplus, the algorithm isn't deterministic; sometimes it takes a second and sometimes it takes 40 :)
#aaronpkgood call on showing the random "working" messages in the browser. makes me know the browser isn't frozen at least
#aquariusit took me one single try of going "gah, has it frozen?" during testing to know that I should do that :)
#aaronpki'm gonna try sending myself a vouched webmention from your blog post
#aaronpkI already got a webmention (pingback) from that bd.summit spam post
#aquariusgo for it. I was going to ask you about that; I did try sending one to one of your posts and it didn't work, but I was halfway through debugging everything and so just put it down to weirdness
#aquariusbut if I'm doing something wrong I'd be interested to know what :)
#aquariusbut it wouldn't surprise me if the tokens die after one minute or after one request or something, and I'd done the endpoint detection once and then tried to send the WM with my script dying about sixteen times, so I assumed that it was just that
#aaronpkoh yeah, my endpoints expire pretty quickly. Maybe I'll increase the timeout so that manual testing is easier
#aaronpk(that was another thought that might prevent some spam, but has yet to be determined if it's effective)
#aquariusit didn't that time because when I write a post I extract the URLs, endpoint-detect on them, stash them, and then run a separate script which WMs everything in the stash list
#aaronpkout of curiousity, do you think that would make any difference in preventing spam? having short-lived webmention endpoints?
#aaronpkcurious to hear your perspective on it, cause you know...
danlyke and verdi_ joined the channel
#aquariusI think it will help a bit. If in order to send you a WM I have to do a request immediately beforehand to get the endpoint, then that's more constraining than building up a huge list of target/endpoint pairs and then giving them to your botnet
#aquariushowever, in general, WM basically *needs* that fetch-endpoint-then-mention cycle anyway, so I have sorta assumed that spammers will just include it
#aquariusbut it *already* makes spam a less attractive target on webmentions
#aaronpki mean the same technique can be applied to pingback
#aquariusbut having endpoints timeout kills half of what remains, because even if you crawl the web to discover pingback endpoints what you *want* to do is put them in a big list and give the big list to the botnet
#aquariusand that doesn't work if the endpoints have vanished
#aquariusthis is why hash-for-vouch is pretty aggressive about making vouch pages disappear :)
#aaronpkah I was wondering about that. I got h-f-v URL #14 so I immediately tried to see what was on #13 :)
#aquariusit is to avoid a spammer who does the 10 seconds of computation to get a page vouched for and then alters that page to have one billion URLs on it and sends vouched WMs from it. :)
#cweiskere skitch: I use shutter together with my shutter scp plugin to publish (annotated) screenshots
#Loqicweiske: tantek left you a message 2 days, 3 hours ago: if you're looking for an opportunity to criticize or harshly double-check h-entry *or* hentry adoption, what *would be* useful is checking to see if anyone *dropped* usage, e.g. check http://microformats.org/wiki/hatom-examples-in-wild#examples_by_category and note which (if any) no longer support it.
#aaronpknice! I don't think shutter works on osx tho right?
aral, LauraJ, lukebrooker and herbsmn joined the channel
#aquariusaaronpk, aha, finally got a chance to try a webmention to your site vouched for by hash-for-vouch and it says, vouch URL is not approved, which is I suppose fair enough, and should hopefully leave it in your database so you can decide whether or not to approve in future :)
#LoqiThe Vouch protocol is an anti-spam extension to Webmention. Webmention with Vouch depends on understanding Webmention http://indiewebcamp.com/Vouch
#aquariuscheck that out. VOuched for by hash-for-vouch. Well done, hash-for-vouch; all your little dreams are achieved. :)
#aaronpkhm, one thought. maybe you could make the vouch URL show a message like "there used to be a vouch here but here's why it's gone now" instead of just 404 not found
#aquariusI note that the URL you link to will disappear, though...
#aquariusthe issue is that I don't *know* that one used to be there, because I delete the database record, so I don't end up with a zillion database records :)
#aquariusanything including the phrase "avoid the database altogether" is pretty much guaranteed to get my interest, but I'm not sure how it's doable in this case :)
#aquariusI, Mr Spammer, create a page with a million webmention-accepting URLs on it, and then do the computation required to make h4v vouch for me, which takes 30 seconds
#aquariusthen I send a million webmentions, all with source:my-million-urls-page
#aquariush4v vouch pages time out in 3 minutes, but you can probably send a million webmentions in three minutes if you've got a botnet.
#aquarius*but* every webmention endpoint will hit the vouch url (well, they will if they care about being vouched for)
#aquariusand so after the first four, the vouhch URL goes away and the other 999,996 URLs don't get spammed because they try to verify the vouch url and it's gone.
#aaronpkI often have more than 4 links in a blog post for example
#aquariushm. 'Cos it was small enough that it gives you a chance to check the vouch url once for paranoia, and then use it, and have a couple of spares.
#aquariusI generate one vouch url per target, not one per source
#aquariusis gtalk still xmpp enough that you can send jabber messages? or did they go full proprietary?
#aquariusI don't really want to run a whole phone app *just* to get notifications of wms to kryogenix.org; even I am not that vain, although I admit I'm close.
#aaronpkthe idea with h-entry of course is that the page would look good to you reading it in a browser, and also is machine-readable for use by other things like the app
#aaronpkthe idea with PuSH is that you can subscribe to my http://aaronparecki.com/mentions page by telling my hub "subscribe to this page and notify me at this URL when something is added"
#aaronpkthen when my site adds a new mention, I ping my hub, which then pings all the subscribers
#aquariusand I understand that it's nice to have one endpoint with data which is both machine-readable and human-readable. But I'm not convinced that I want APIs to start returning HTML that I have to parse rather than JSON :P
#GWGThis is a job for the WordPress Outreach Club.
#aaronpknobody parses JSON by hand. they use a library which parses the text and turns it into native data structures. nobody parses HTML by hand either. they use an mf2 parser which turns it into native data structures
#aquariusJSON.parse(string). Yep, that's hard work, but someone else has already done it for me. Having to include a library is pretty annoying, especially if I'm doing it client side :)
#tantekfeels compelled to say the needs of the publishers outweigh the needs of the parsers.
cmhobbs joined the channel
#aquariustantek: I *broadly* agree (that the needs of the publishers outweigh the needs of the parsers; more importantly the needs of the *consumers* outweigh the needs of parsers, no?), and there's certainly some merit in basically saying "away with machine APIs! let us always use human-readable data with appropriate markup! no JSON for us!". But that's a reasonably hard sell, especially since we've been talking about it
#aquariusfor years and basically nobody does it. :)