jbenet
commented
9 years ago thanks @ChristopherA !
there's a ton to say on this.
the nym idea is solid! nyms would be particularly useful for ad-hoc ephemeral discovery, in the etcd style.
nyms aren't globally unique (they're not meant to be). it may be easy to extend to using other naming systems (like DNS, keybase, etc) for global identity.
I'll also add in the SFS ideas. Relevant papers here: http://gateway.ipfs.io/ipfs/QmR9yzggUjRQRWBuDgwwjrTWH1Vhd37U8PxNNLYoCN5xYv -- Section 3.4 in the phd thesis one (beware, the pdf is large...).
Relevant also to think git style "remotes" -- which is a way of binding a remote node to a local name. We could have keys though, which is safer :). this would work kind of like a local IPNS symlink. This is functionally the converse of a nym, where it is assigned by a local entity. but this is more in line with SFS certification paths.
/ipns/remotes/whyrusleeping/...(lots of half-formed thoughts. figured i'd drop a few here for now and come back later)
ChristopherA
I recommend that you add an optional pseudonym (aka nym) to the
ipsf daemonthat adds an non-unique identifier to theipfs peersoutput.Initially this aids debugging the swarm discovery process when demoing ipfs. Longer term it can bootstrap some experiments with SDSI-like local names i.e. ChristopherA's JuanBenet
For the initial prototype, you may want to limit displaying the nym to only the first 20 characters so that 120 character wide terminal displays do not wrap. You may want to limit the possible characters to ASCII upper/lower and a few limited characters (to allow @ChristopherA, ChristopherA@gmail.com, etc.), or even to not allow special characters and keep it to twitter name restrictions (15 characters upper lower plus _).