164800+ entries in 0.101s
mircea_popescu: switch over my shoulder grinning and bobbing her head. "yeah, what men need
to perform is constriction."
a111: Logged on 2017-08-22 14:35 asciilifeform: when saw
that
the ~original~ src was a pile of shithacks, lost interest in anything but
the electron microscopy path ( like it or not, 1uM process folks were ~forced~
to make compact description )
mircea_popescu: not a bad idea at all ; was going
to come up when we were finally making
the
tmsr hdd controllers. but even early dun hurt anything.
a111: Logged on 2017-08-22 14:20 mod6:
http://btcbase.org/log/2017-08-22#1701801 << Sounds like
there should be a link
to
the
trusted-nodes page in
the HOWTO maybe. Also, a once per-month round-up of me asking for Node-Updates, if
there are any.
mircea_popescu: asciilifeform yeah, which aspect makes
them slightly better
than spinsters, which actually become unreadable.
mircea_popescu: (raid reconstruct is io intensive, could push over
the edge
the redundancy, dying disk)
mircea_popescu: as age is a great predictor of ssd failure and
the shits are perfectly capable of dying same week.
mircea_popescu: but yes,
the principle is correct : make raid out of same items,
then a few months in change one.
though it doesn't need
to be changed.
then use
THAT in
the next raid you build.
a111: Logged on 2017-08-22 12:40 asciilifeform: say you have 4 holes. use 3 mechanicals + 1 ssd,
then few months in, replace a mechanical,
then again, year later, whole
thing is ssd
that will not ever simultaneously burn , in
theory.
mircea_popescu: it's important
to find out, after all most fiat unis/tech corps/whatever actually
to
this day harbor
the managerial delusion
that
they can in fact compete with
the republic on a flesh basis.
mircea_popescu: i don't specifically care. it's a simple "come see whether you are good enough
to seep people into your company or lose all your brain power
to our better model".
mircea_popescu: that's
the fucking position, wtf do i want FROM a bunch of fiat rottinculo.
mircea_popescu: not a matter of
that. a matter of, hey, we're actually significantly smarter
than you, come hang out, who knows, maybe you gain something.
shinohai is prod
to assert
that he knows a woman irl skilled in
the handling of wartenberg wheel
a111: Logged on 2017-08-22 12:25 asciilifeform: but lattice per se is EXACTLY like xilinx, same profit model, closed arch, license 'ip cores'.
their larger flagship fpga is exactly like xilinx 'spartan', full of proprietary peripherals, and
that's
the one
that
tends
to get packaged into devboards with nic etc
mircea_popescu: we should get ice-cube
trays with various
TMSR ship names on
the inside.
mircea_popescu: there is no substantial difference between us and british navy just like
there's no substantial difference between us and nazy atomic program.
mircea_popescu: wait, is it because india names its ships NSS Blabla sopmething we're supposed
to believe "that's just what';s done" ? rather
than "oh look, orc HMS!!!" ?
shinohai: And I
thank
thee, mircea_popescu , fpr reminding me
that
the US Navy + Pykrete = eterenal meme
a111: Logged on 2017-08-22 12:08 asciilifeform: ( implementation becomes an underhanded-C-contest in concealing
the fact of ~any~ box running
the idiocy reverting
to nullcipher on demand )
a111: Logged on 2017-08-22 10:36 shinohai:
http://archive.is/0TOaA "The US Navy orders "Operational pause" as it
teaches sailors
to actually navigate waters and use GPS, whilst all llitoral combat ships are refitted with pykrete
a111: Logged on 2017-08-22 10:17 valentinbuza: spyked, people who are serious about
transport security (data in
transit) shy away from
TLS and
they craft
their own stripped down version using Noise Protocol Framework (
http://noiseprotocol.org/index.html) spyked: shellshock: "let's call
this general-purpose function
that executes programs in a shell".
spyked: yes, and C is I
think it's a good example
to illustrate
the larger issue. it's a snowball
thing, in
the sense
that it's sometimes enough
to have 1 hole
to break everything. incidentally most recent popularized vulns (not necessarily in C) fit
there.
spyked: think about it,
the problem of e.g. C software is
that unsanitized inputs let users do *whatever*
they like with it, i.e. arbitrary computetion. which goes way beyond program specification.
spyked: valentinbuza, maybe not, but
then if you have everything loaded in head,
the most you can do is rip
the useless parts apart and leave *only* what fits into
the problem at hand. which
turns "framework" into "item
that solves particular problem". it is essential
to not leave *anything else*
there.
valentinbuza: 'time
to learn,
the "framework" becomes useless, because
the mental framework is in place.' << my guess is not, but I don't
think we have a conclusion on
this with a sample of 2 points
spyked: on
the other hand, if I
take
these items for granted, joke's on me, which is exactly what "modern engineering" philosophy relies on.
spyked: loaded (because I don't use
this day-to-day), I must consult
these items in great detail. my guess here is
that once I have spent all
the
time
to learn,
the "framework" becomes useless, because
the mental framework is in place.
☟︎ spyked: valentinbuza,
to exemplify asciilifeform's point ^ I shall quote from
the docs: "A Noise protocol begins with
two parties exchanging handshake messages. During
this handshake phase
the parties exchange DH public keys and perform a sequence of DH operations" <--
this requires me
to import a couple of concepts: handshake messages, DH public keys,
there may be others along
the line. now, given
that my crypto brain-memory module is not
valentinbuza: asciilifeform,
two
things can suck and one can suck less. But instead of
throwing a lot of arguments, why not propose your ingredients and recipes?
spyked: re schematic for protocol patterns, why not use e.g. petri nets for
the model (assuming
that works)
then just implement from
that? why add extra software? ehm. it seems like
they're
trying
to automate some work, but
that automation
trades off actual understanding, i.e. by introducing (IMHO useless) levels of abstraction.
valentinbuza: probably. but i
think
that your argument is invalid because you say
that "in
TLS ingredients suck and recipe sucks" and "in Noise ingredients suck
therefore
the recipe also sucks"
☟︎ valentinbuza: it's an attempt
to make some
things better
than
TLS (or other data in
transit protocol) as opposed
to other ways of creating software such as "we use
TLS because it's standard, we have no clue what
to do and just use what everybody is using" and sell it as military grade
valentinbuza: i don't find
the word 'standard' in
the description or in
the spec. it's not a standard and should not be seen as one
valentinbuza: probably
the word framework is misunderstood. Let's say you want NOISE_CURVE25519_ETC it does not provide you with curve25519 implementation, you have
to create you own. It's just a schematic for protocol patterns, not a framework a la "django"
spyked: valentinbuza, my issue is
that "framework" approach (as used in
today's
terminology) is utterly anti-engineering. one can (on
the condition
that
they know what
they're doing! and
there really is no alternative
to
that) write own software from first principles without requiring 3rd party. or, use 3rd party only
to strip of shit and output sane object (e.g.
http://btcbase.org/log-search?q=trb ), which is distinct from "framework".
valentinbuza: as you can see on
the spec, it is not concerned with PKI or your authentication methods, it's up
to you
☟︎ valentinbuza: agree on
the
TLS part. As I
told before, Noise was a partial response for spyked blog post (TLS sucks, PKI sucks). Noise is just a somewhat better choice for
the
TLS sucks part
valentinbuza: also "massive pile of moving parts" << not even close
to
TLS. as for your other questions i can't really answer.
valentinbuza: don't know. ask
Trevor Perrin, maybe he
thought of creating all
the possible recipes
☟︎ valentinbuza: it is different from
TLS, where whatever version you are using it has null cipher.
The question should be: does someone deployed NOISE_NULL_CIPHER_TOTAL_BS?
then you can blame
them
valentinbuza: noise is a framework for creating protocols. you have
the option
to create NOISE_NULL_CIPHER_TOTAL_BS protocol which is
totally different from NOISE_ANOTHER_SANE_CHOICE
☟︎