#snarfedthat's what i mean. same issue except github instead of twitter
#aaronpkyes that comment accurately describes the problem
pasevin_ joined the channel
#aaronpkif a site serves the same profile from two different URLs, that's a problem
#aaronpk"trailing slash" is not really a thing, the URLs are different
#snarfedyes, true, but it'd be reasonable to treat the same url with and without trailing slash as identical
#aaronpkthat involves hard-coding stuff to providers, which I could do but that's annoyiung
paulcp_ joined the channel
#snarfedannoying for you, yes, but friendly for users :P
#aaronpki can't just assume all links with a trailing slash are identical to the non-slash version
tantek joined the channel
#snarfedi guess. the provider list is already hard-coded, though (right?), so blacklisting or whitelisting doesn't seem too hard...?
#snarfedanyway. just a thought. let me know if you want me to file an issue
#aaronpksure go ahead, would be good to have a list of providers which have this behavior
#aaronpkwhat github and twitter should be doing is sending a 301 redirect from twitter.com/aaronpk/ to twitter.com/aaronpk and then everything would be working fine
#snarfeddefinitely! agreed. but they don't. :/ ah well.
#kbssnarfed: for the four remaining people who actually use pgp for contact info :) think its worth removing old email addresses from https://snarfed.org/pubkey.txt ?
#kylewmaaronpk, it's really testing your facepile arrangement :) looks good
#tantekkbs, snarfed which reminds me, how does one *obsolete* an email address (or other contact info) in the context of PGP contact info (or frankly any contact info) ?
#kbssnarfed: oh, and if you link to it from your home page
#kbswill make it easier for me to discover ;) just a <link rel="key" href=""> or something similar...
#tanteklike - "I am no longer using this phone # ..."
#tantekthe use case being - "dear friends, please delete this number you have for me from your contacts/addressbook"
#kbstantek: unfortunately - it ends up being (in practice) what any given client does. In theory, a client with a pgp key ought to be checking the keyservers for updated (or revoked) information
#kbsand you can update (or obsolete) information by revoking an earlier signature
glennjones joined the channel
#tantekkbs - so you revoke a signature and that revokes *all* the information as part of that signature?
#kbspretty much. [You could also use u-key in your hcard, if you want to make it visible in the html as well]
#tanteksnarfed - I thought rel="key" was derived from links on the web
ttepasse joined the channel
#kbstantek: you can choose to revoke a specific portion of the key file. Each uid (== a particular email address) has its own signature in the key. You could just revoke that one signature alone
#tanteksnarfed, n.m. looks like none of the examples actually use rel=key - but could
#kbsso it's big and painfully complicated. There are internal references to information like bits of text (like an email) or subkeys that are validated by signing each reference (more or less)
#kbsand finally, the revoking mechanism allows you to tell a keyserver to delete a prior reference - perhaphs it's easier to think of it like a version controlled system, where the end result is the result of combining all the intermediate 'changes' to the repository
#snarfedben_thatmustbeme: interesting. minor point:, you return 200 for urls that don't exist and serve the home page, e.g. http://ben.thatmustbe.me/asdf
#ben_thatmustbemeyeah, i haven't dealt with that yet. I noticed that
#ben_thatmustbemefor the site I had originally hacked this up for, they wanted that (much older crowd was the user base, so basically just assume that typos are probable)
#snarfedheh. you can return status code 404 and still render the home page
#tantekso when canonicalizing profile URLs from a paticular provider, if that bit was set, the could would check and drop trailing slashes on profile URLs of that provider (regardless of source)
#tantekso regarding this SSL exploit - does this mean stop using various websites for a while?
#aaronpkall I know is that the exploit can allow attackers to dump parts of the system memory
#aaronpkhow that affects me feeling safe about signing in somewhere I am not sure
#tantekhow do we know which providers have upgraded their SSL?
#aaronpkkbs: fun fact: it can be a criminal offense to do that
#kbsThe potential issue is also whether someone has logged into a site over the last day or so - what's also happening is that the credentials of people who've logged into a given server is available in the dump.
#Loqikbs meant to say: The potential issue is also whether someone has recently logged into a site over the last day or so - what's also happening is that the credentials of people who've recently logged into a given server is available in the dump.
#kbsold-fart-anecdote - I was running a crawler for my thesis that would ping random SunRPC services on the internet to determine their uptimes. Bug in some of the services caused their computers to crash, hacking suspicions - much drama. fortunately was able to deflect most of it to my advisor...
_skinny joined the channel
#kbsoh, the times when there was no such thing as a "firewall" :)
ireheart, sparverius and pasevin joined the channel
#aaronpkrebooting the server real quick with new openssl
#aaronpkcan't remember at the moment, it came up during indiewebcampsf
#aaronpktantek: I just got an iPod for testing stuff at work, I hadn't actually seen one of the new ones yet! Is that the one you have? the 4" retina one?
#ben_thatmustbeme!tell damn you barnabywalters, by asking me a question in your webmention you are fueling my desire to have sending working that much more.
#ben_thatmustbemeat least people can see that I know they mentioned me. I just know I'm going to have a ton of stuff to try and strip out any javascript or funky html to make sure its safe before i show it on my site at all
#aaronpkben_thatmustbeme: you could start with just showing the plaintext version
#ben_thatmustbemestrip out any tags, that should be all i'd need i suppose right
#aaronpkben_thatmustbeme: you're welcome to try to craft an attack and send me a webmention
#ben_thatmustbemeit would be an attack on someone else that mentioned you already, and resubmitting that mention on my own with a URL extended to include JS, Its very unlikely to find someone that has a site like that, but I may set one up just to test feasibility
#kbsyou might be able to just use paste.debian.net I think
#aaronpkwe should make a webmention vulnerability test suite, hehe
#aaronpkwebpwn.com/hack/my/site?me=aaronparecki.com <-- generates an attacker page linking to a post on my site so I can send webmentions from that URL to test what happens
#aaronpkit may not fit into micropub since micropub assumes it'll be creating content...
#kylewmso actually would there be any advantage of sending source=bridgy-url&target=kylewm.com/original_post%3Fsyndication%3Dtwitter_url as opposed to just notifying the user's homepage
#kylewmhaving a (link) at the end of the POSSEd tweet is good in that it { cites original content, serves as micro-evangelism } and bad in that it { is a little distracting, uses up characters }
#kylewmdo folks generally feel like the benefits outweigh the costs
#snarfedi hadn't thought much about how to preserve backfeed. query endpoint, extra "syndicated" webmention param, searching h-feed entries for rel-syndication are all possibilities
#gRegor`kylewm: I haven't implemented POSSE yet, but I've not been a big fan of adding the short link at the end of the tweets. I think I would definitely like an alternative.
#aaronpkkeep in mind there are also human benefits to having the link, like if visitors can click it to find your site where there is a *better* experience than on twitter
#gRegor`Though I feel like it would be obnoxious to click through and see nothing additional / new. E.g. no interactions.
#kylewmaaronpk, have you had anyone complain about the close ) being included in the URL when they click it? I dont know if it's tweetdeck or what that does that
#Loqiaaronpk meant to say: gRegor`: that is the negative feedback tantek talks about, which is totally justified if there is no additional content
#kylewmsomebody mentioned that to me but i didn't get details
#gRegor`I feel like it also might lessen interest in clicking my non-permashortlinks. Like when I'm sharing a link to an external site I find interesting.
#kylewmgRegor`, yeah I totally have that concern too, or at least make it confusing which link they should click on
#gRegor`I guess a youtu.be link would be obviously different from my own shortlink, but wonder if people would get used to overlooking my links. :)
#gRegor`I'll worry about that once I get POSSE going, though. Heh
#kylewmi've noticed barnabywalters does \n(shortlink) ... that's pretty inobstrusive
#gRegor`!tell KartikPrabhu Still on for tomorrow night?
#tantekbarnabywalters: was just about to say waterpigs.co.uk/pushups/ is not loading for me and then I saw last several lines - sorry to hear about that.
#tantekis it on github? I'll file the same feature request issues that I did with aaronpk
#barnabywalterstantek: not on gh yet, I can put it there if there’s interest
#tantekbarnabywalters - yes! definitely. I'd love to see a webapp version of a pushups app
#tantekand happy to use github issues to send you feature requests too
#pauloppenheimkbs - reading logs from... 7am?!? root certs, if online (which they shouldn't be) would need to be changed. Most are offline though (i dearly hope that's a req of having a root cert)
#kbspauloppenheim: makes sense - (think cweiske also indicated as much)
#pauloppenheimkbs: i run an intranet CA, and the private key is offline and on a machine that has never connected to the internet.
#kbspauloppenheim: nice. Yes, I was just wondering aloud - as you say, would only matter if the private key was ever actually directly used on a server and that'd be rather terrible for a root CA to do.
#pauloppenheimkbs: well, they all have child CAs for doing the actual signing work, which possibly *are* online, so that'll probably wobble over the next few days
#pauloppenheimKevinMarks_: cert pinning helps other problems, but AFAICT not heartbleed. If the server is using openSSL 1.0.1 and has a TLS heartbeat, it was vulnerable to having memory contents read, which would include any private key.
#KevinMarks_right, I meant that cert pinning is going to make it harder to replace all the certs everyehere
#pauloppenheimhence the server needs to make a new private key, and you need to get it to pin it.
#pauloppenheimoh, possibly, depending on how one pins their certs