log☇︎
5000+ entries in 0.019s
Framedragger: asciilifeform: does your script pull out only these debian-bullet-fucked moduli? would be nice to be able to distinguish in the-db-which-im-gonna-put-them-in
Framedragger afk for a bit
Framedragger: yes correct
Framedragger: i hear ya
Framedragger: (i hope they're in a decently parseable format, but i think they are)
Framedragger: okay sounds good, will do!
Framedragger: i can take a look at the data to see how easy would it be to stitch things up next ~wed
Framedragger: yes
Framedragger: but yah.
Framedragger: it's good amusement tho
Framedragger: :)
Framedragger: (well, so the story goes of course. either way, malice / insanity)
Framedragger: mircea_popescu may be interested to read about how it happened. i think it was some developer-helper tool / linter / sth which showed an 'unused var'? and dev removed it. and broke rng
Framedragger: my gerd still so many of them in there
Framedragger: ahh, i missed the qntra news item.
Framedragger: https://www.reddit.com/r/btc/comments/5ddiqw/im_haipo_yang_founder_and_ceo_of_viabtc_ask_me/da3rg4q/ << konspiracy-by-having-a-different-opinion-on-bitcoin :p
Framedragger: thestringpuller: you mean this? https://twitter.com/ViaBTC/status/797160809306714112?lang=en
Framedragger: http://m.bvulpes.com/b@bvulpes.com << cool stuff
Framedragger: though, e==35 (not prime) in these it appears, so prolly not.
Framedragger: ye olde debian prng bug => small set of possible ssh keys => factors extracted for keys => factors inserted into db?
Framedragger: http://log.mkj.lt/trilema/20161117/#94 << heh.
Framedragger: nicht so schlecht
Framedragger: right.
Framedragger: "only a matter of time", i guess
Framedragger: trilema.com does not seem to expire
Framedragger: (if no trust in domains in the first place, what difference if 2nd level or 3rd level)
Framedragger: mircea_popescu: i'm curious, why don't you do mpex.trilema.com ? or would that link the two explicitly and hence would not be favourable?
Framedragger: 5518 * the NS RRset is allowed (Type 4).
Framedragger: 5517 * rcode is NXDOMAIN (handled above) in which case
Framedragger: 5516 * a SOA or a NS RRset (Type 3, handled above) or
Framedragger: 5515 * and a optional NS RRset (Type 1) or it has neither
Framedragger: A negative response has a SOA record (Type 2)
Framedragger: i also like such Brussels born comments as
Framedragger: trinque: oh god i need to delve into a longer and madder thing it seems https://source.isc.org/cgi-bin/gitweb.cgi?p=bind9.git;a=blob;f=lib/dns/resolver.c;h=74d82ae1076a80dc8edf8964b295b40d93b735d1;hb=4e8fe357a619ae2b789b15df7e3d6abf782b4a71 - goodtimes
Framedragger: (..and of course bind has at least one "basically completely undocumented" config option, "order random_1", which modifies the round robin process, or the eventual reply. "didn't you get the memo, peter?")
Framedragger: BingoBoingo exactly
Framedragger: array resizing and network code all in one place. no wonder buffer exploitation bugs aplenty
Framedragger: yeah....
Framedragger: trinque: thank you! will check.
Framedragger: mircea_popescu: oh you can't even imagine how many 'additional' data fields domain needs to be actually usable by current dns clients. everything from start of authority to multiple TTLs etc. and that's the saner part of things...
Framedragger: s/necessary// ^
Framedragger: yes indeed..
Framedragger: and here i thought i could exploit other people's knowledge :p
Framedragger: the working assumption would be that if necessary does permute, it will necessarily set a new order for each request to follow. dangerous assumption, sure.
Framedragger: can this be used to approximate the frequency of dns requests to that nameserver? :p
Framedragger: curious: when looking at dns round robin, i found out that some nameservers - of those who do round robin, i.e. permuting list of say dns A records - sometimes do not shuffle them around, if dig'd in quick succession.
Framedragger: :(
Framedragger: 'GO TO WORK'
Framedragger: already sent to some 'trump is evil' friends
Framedragger: i also liked the part where if javascript is given a file and it doesn't recognize if it's a js file, it'll just try to eval it, for good measure, in case it is js.
Framedragger: http://btcbase.org/log/2016-11-16#1568229 << ikr - and yes any downloaded-from-google-and-imported lib can do this... ☝︎
Framedragger: so the conflict thing is, i suppose, in principle settled in my mind.
Framedragger: http://btcbase.org/log/2016-11-16#1568065 << i guess http://btcbase.org/log/2016-11-16#1568062 explains it - there are no 'other rounds'. before-sunrise, and after-sunrise only. ☝︎☝︎
Framedragger: mornin'
Framedragger: bbl, sleep
Framedragger: first round, okay, but other rounds supposedly follow later, and the conflict case can be replicated, i would think, and hence the system supports conflicts, structurally, which need to be addressed one way or another.. dunno. ☟︎
Framedragger: because it would be great if WoT-gns worked like other WoT things, i.e., via the enforcement of WoT itself. ☟︎
Framedragger: and mp said "da fuck, I'm not going to support a dns system where some dork registered trilema", which confused me, so yeah, clarification needed, i think.
Framedragger: pretending that they are MP?"
Framedragger: http://btcbase.org/log/2016-11-15#1568014 << that's what i thought as well - i queried along the same lines: "say someone with a valid GPG key rushes to register trilema.com in the Republican DNS before yourself. I suppose that is all well and good, and you negrating the key would only be appropriate in the instance of that person ☝︎
Framedragger bbl, too
Framedragger: also, http://log.mkj.lt/trilema/20161115/#622 << http://btcbase.org/log/2016-11-13#1566745 but this may be an issue in semantics/definitions only, i suppose. ☝︎
Framedragger: but then already it is not as simplistic as "whoever owns the key", no?
Framedragger: unless the latter kind of situation is to be covered by your clarification, to quote, "I imagine like any serious country, we first handle the claims of the elite privately. " ☟︎☟︎
Framedragger: will need to change places in a few min, too.
Framedragger: mircea_popescu: i (or someone more credible, or whoever) registers "trilema.com" under their own leigt key. what nao?
Framedragger: mircea_popescu: i guess in your view, perhaps no viable conflicts would arise in gns? say, two owners of two distinct pgp keys claiming ownership of "apple.com". some kind of due process is to take place, presumably
Framedragger: right. how would the process of resolving conflicts in gns look like, though? (i'm just curious, i.e. question is well-intentioned, not troll-y)
Framedragger: i guess that's a given assumption in #trilema
Framedragger: however, an agreed-by-everyone-who-matters symbol context would still be needed. (sorry if interrupting, trying to clarify for my own education.)
Framedragger: so the notion of a local symbol frame/context will be retained inevitably, is that what you're saying mircea_popescu? (in which case i'd add that gns-the-implementation could even probably be used - on each interested user's machine - as a local name system. etc.)
Framedragger: i still have a reserved name ('indra', as in reflexive "indra's net" / indrajāla) for my mega decentralized permanent content concept, baked when i was maybe 18 yo lol. in fairness, "indra's net" is a l33t name that i hope i will use some time.
Framedragger: aha.
Framedragger: is my (naive probably) understanding.
Framedragger: could be a set of particular (and particular only) peers. in any case, there must be a shared understanding among the users/elite as to how to proceed in face of naming conflicts.
Framedragger: "nexus of hierarchy" connects for me (maybe on some superficial level only) to kyristor's "The only questions A0 is qualified to answer are those for which disputes can be settled entirely through majority vote of the entire DHT network" - where the dht network
Framedragger: "A0 can be implemented by regarding an authority's PGP public key as being its public routing address!" << nice kademlia and/or gossipd vibes
Framedragger: i guess counterpoint would be http://log.mkj.lt/trilema/20161115/#456
Framedragger: asciilifeform: how about: make a proof of concept name system, use it instead of current dns root server set for now, later enable every gossipd user to run their own instance of name system if they prefer the fully-decentralized-dictionary path; the initial PoC will still have been useful.
Framedragger: asciilifeform: i thought so too, but basically mircea_popescu's idea would dispatch of the whole 'dns zone' concept. there would be no 'domain' per se. dns clients could still query 'loper-os', but the server would be a simple table, with no understanding of zones or significance of "."
Framedragger: personally i don't see why there could not be a GNS which would be separate from commitments to specific transport standards. have a table, have a transport layer, swap the layer later. i may be naive in regards to the "swap" step, i guess... :/
Framedragger: but i speculate.
Framedragger: there could be WoT members maintaining their own namespaces that others may want to peruse, etc.
Framedragger: so, yeah. and i'm no longer convinced it would redundant, in the sense that when gossipd cometh, one must throweth the 'general name system' away
Framedragger: "allegedly"
Framedragger: i don't have a strong opinion, i wonder what mircea_popescu thinks. i guess the answer would also focus on the "while we still must use it" aspect
Framedragger: i see what you mean.
Framedragger: trinque: right, this is cf. gossipd's "everyone has their own a la hosts file, and does with it what they like". is that what you meant?
Framedragger: trinque: i meant as a generic string - trying to avoid the term 'domain' as the latter is not accurate..
Framedragger: trinque: but yes, the central table thing remains
Framedragger: trinque: not that part. the transport part - name query/response, as well as name updates.
Framedragger: re. plaintext for all NSAs to read, sure, there's that. goes against gossipd's "no free bits for the unauthenticated" i guess. but again, this can be wrapped in gossipd later.
Framedragger: http://log.mkj.lt/trilema/20161115/#440 << sure, i guess. (note though that this effort would get rid of dns server code, though. and it does not obstruct one from later patching dns client code / rewriting a much more simplistic name query client.)
Framedragger: right, both precision and accuracy required (any error rate cripples the UX, etc.)
Framedragger: ah! k
Framedragger: asciilifeform: "it was interesting learning experience in that i never understood how ~motherfucking reliable~ an ordinary keyboard must be before it feels usable." very interesting! nice project. (ppm == pixels per meter?)
Framedragger: asciilifeform: which is, i think, not a complete retreat, so to speak
Framedragger: asciilifeform: do note that mircea_popescu's idea of keeping dns is more akin to a general WoT-enforced hashtable, update-able via (in principle) gossipd-compatible pgprams, and (for the time being) transportable over dns/udp. the latter so that dns clients can make use of it.
Framedragger: (in fairness, the html sometimes mangles two words together, omitting a space, or somesuch. not many words are lost. but still, shame.)
Framedragger: (and yes, agree.)