#hypothes.is

/

      • woah
        I want to encypt messages with a public key
      • propagate them across the network
      • and have the node with the right private key read them
      • it would be nice if i could avoid some kind of nonce exchange scenario
      • maybe pgp is better?dunno
      • tilgovi
        ah
      • well
      • if an attacker can't predict what's in the message
      • and the effect is idempotent
      • and doesn't "grant" anything to the sender, other than that the receiver processes the message
      • then I think you wouldn't need a nonce
      • idempotent + no "access" granted to the sender, no "session" of any kind or anything given in return for the message, would mean there's no threat of replay
      • and if an attacker can't predict the contents of the message, or cause arbitrary messages to be encrypted and sent, then there's no chosen plain text attack
      • woah
        yep, sounds about right
      • so for more background on what i'm building
      • tilgovi
        normal asymmetric algorithms, public key cryptography, should do just fine
      • If there's no harm in the receiver getting a message a second time, from someone other than the sender, then I think you've got no need.
      • woah
        this thing produces chains of messages, each containing the hash of the previous message
      • and a squence number
      • tilgovi
        sequence number is a great nonce
      • might as well use it :-D
      • woah
        yea true, it is public tho, but i guess it doesn't matter
      • tilgovi
        doesn't matter at all
      • you could use it like a high water mark
      • rather than remember every nonce to be sure it isn't used
      • woah
        how?
      • tilgovi
        just remember the last one and reject any message with a nonce <= the high mark
      • woah
        oh yea
      • tilgovi
        you've effectively ensured that each one is only used once
      • woah
        that's part of the main thing anyway
      • tilgovi
        consensus?
      • woah
        nah actually not
      • no global consensus
      • this is based on secure-scuttlebutt which i may have told you about but it is a bit different
      • tilgovi
        Okay. I thought it might be a consensus algorithm, cause we were talking lots about those recently and paxos (I imagine raft, too) has this concept of a promise never to accept lower proposals
      • wait... you're gossipping these messages to get them from sender to receiver yeah?
      • woah
        that's not really specified
      • tilgovi
        okay
      • woah
        i'd like to do something where nodes pay cryptocurrency for transit, but that's in the future
      • tilgovi
        cause I was going to say if other parties are forwarding and replaying your messages by design, nonce makes no sense.
      • :-D
      • woah
        yea i think i should just disregard it
      • tilgovi
        "Please deposit .004 BTC to continue"
      • woah
        you pay for internet access right?
      • tilgovi
        OMG ROUTING FRAUD
      • woah
        ROUTING FRAUD?
      • tilgovi
        I'm imagining operators forwarding packets along non-optimal routes to defraud the sender.
      • chrisbirk has quit
      • woah
        yea we talked about this stuff at gil i think
      • tilgovi
        never mind, just getting excited
      • I don't remember that coming up specifically, but yeah, we talked a lot
      • woah
        yea that would be a big issue
      • im just starting small
      • tilgovi
        anyway, I should maybe get back to figuring out how we broke our chrome pdf support :-(
      • woah
        oh man i hate pdf
      • in sympathy
      • tilgovi
        looks like I have bigger issues
      • dwhly: I'm looking into it
      • Strange that I didn't catch this before I uploaded it and I can't reproduce it locally.
      • dwhly
        tx tilgovi
      • tilgovi
        okay. no problems with a local dev chrome extension either
      • trying the prod extension
      • that fails
      • hm
      • minification issues?
      • dwhly
        tilgovi: it strikes me that the issue w/ the pdf reported on the forum may have to do w/ that particular one-- it being hosted on github.io w/ an .html suffix or something. seems adder doesn't show on that one, but i'm not seeing that elsewhere.
      • tilgovi
        you can annotate other pdfs?
      • dwhly
        yes.
      • this one
      • it's my usual test pdf
      • tilgovi
        OH
      • that's not a PDF file
      • you're right
      • dwhly
        that's where i noticed the other bug
      • tilgovi
        that's the PDFjs viewer
      • dwhly
        yep
      • tilgovi
        got it
      • :sweat_smile:
      • GitHub71 joined the channel
      • NOTICE: [h] tilgovi pushed 1 new commit to master: https://github.com/hypothesis/h/commit/eb3c486991d921a1f2342a289cb9755d9d76459e
      • NOTICE: h/master eb3c486 Randall Leeds: Show privacy dropdown even on empty permissions...
      • GitHub71 has left the channel
      • GitHub2 joined the channel
      • NOTICE: [h] tilgovi pushed 1 new commit to master: https://github.com/hypothesis/h/commit/7bd01339fdfd5972a809fabd83925dee54705eb1
      • NOTICE: h/master 7bd0133 Randall Leeds: Default visibility private
      • GitHub2 has left the channel
      • okay
      • that should fix it
      • now I gotta get this out, kill time until it propagates
      • and then go fix the permissions on any annotations people made in the last hour or so
      • but, I introduced this bug
      • so... I suffer
      • what's unfortunate is we added a bunch of tests for this code
      • and the tests didn't catch this case
      • well
      • some of it was me
      • and the part that I messed up, tests did not catch :(
      • other part would have required integration tests and all we had was unit tests
      • pushed and published
      • i'll wait for that to propagate before I go fixing permissions
      • and write some more tests right now
      • maybe one of these days we'll have an uneventful deploy ;(
      • Okay. I think was allll my fault.
      • dwhly
        tx for tackling that tilgovi :)
      • tilgovi
        dwhly: np
      • sorry I caused it
      • GitHub22 joined the channel
      • NOTICE: [h] tilgovi pushed 4 new commits to master: https://github.com/hypothesis/h/compare/7bd01339fdfd...21ebfcd5665c
      • NOTICE: h/master c8c62b3 Randall Leeds: Revert "Show privacy dropdown even on empty permissions"...
      • NOTICE: h/master 93ce1e1 Randall Leeds: Support memory storage removeKey
      • NOTICE: h/master 0fdbba6 Randall Leeds: whitespace
      • GitHub22 has left the channel
      • glad I deployed in the middle of the night
      • and then didn't go to bed like I should have
      • dwhly
        we might consider going back to the staging extension as an alternative (not sure what was involved in this deploy)... at least to give it a couple hours shakedown.
      • leathan joined the channel
      • tilgovi
        dwhly: yeah. I'll make a plan for what to do with staging.
      • I pretty much discontinued it for the moment.
      • btw, there were 6 annotations made in the window of brokenness
      • one was the person on our user forum, how fortunate
      • so I'll respond there
      • one was you
      • one was me
      • dwhly
        K
      • tilgovi
        actually
      • we were the only three people active
      • and I fixed the permissions
      • you sholud be able to edit/delete etc now
      • dwhly
        Ok. Will do.
      • tilgovi
        well
      • when the extension updates
      • any moment now
      • woah
        I think you guys need more prominent labels on the little icons that drop down from the menu bar speech bubble icon
      • tilgovi
        dwhly: all set
      • dwhly
        Tx
      • tilgovi
        woah: i'm going to bed, if you can file an issue for the broken link and where it appears that'd be great
      • night
      • tilgovi has quit
      • dwhly
        Woah: you mean like something other than tooltips? Or more rapid tooltips?
      • MrWoohoo joined the channel
      • woah
        dwhly just something that is part of the ui, that doesn't move. maybe to the left of the icon
      • rather, it moves along with the icon
      • woah has quit
      • Sebastien-L joined the channel
      • csillag1 joined the channel
      • GitHub96 joined the channel
      • NOTICE: [h] csillag created more-consistent-names (+1 new commit): https://github.com/hypothesis/h/commit/267ea1959800