257200+ entries in 0.174s

ben_vulpes: phf: is
that particular 'compact size' anything like other hand-rolled variable length integers in c-land?
phf: well, step
two after writing compact size reader/writer is
to figure out how
to make a general purpose "binary-type object of count `compact size`"
ben_vulpes: noty, i would like
to constrain
the attack surface
to satoshis retardation.
ben_vulpes: otherwise what, write parsing of entire structure by hand? and
then reserialization?
ben_vulpes: asciilifeform: it is imho entirely worth
the sweat
☟︎ ben_vulpes: conveniently,
the generic for read-binary accepts other keys, into which i can pass a length for reading
phf: the writer likewise knows all it needs
to know about how
to serialize from ~the number~.
the check becomes "in which range it is, in which case write it
thus"
phf: so it's up
to reader
to decide if it should return first octet as ~the number~ or read first octet and read a bunch of stuff after and return
that as ~the number~
phf: the
type is called compact size (in bitcoin src parlance), it is simply an number.
the reader reads a byte, decides what
to do, ultimately returns ~the number~
ben_vulpes: phf: i do believe
that i have
to hold onto first-octet, if it's less
than 253 it *is*
the length.
phf: well, while asciilifeform's yak shaving, i don't
think
this is correct way
to handle
the
type
phf: ben_vulpes: you don't have
to store first-octet since, it's a property of variable-integer (also if you change variable-integer, you'll have
to make sure
to correspondingly update first-octet)
ben_vulpes: funny,
that's
the exact same misread
trinque made.
ben_vulpes: i wasted an inordinate amount of
time on various false starts in storing variable length fields.
ben_vulpes: (if
the
thing had an adult database,
the
tree would fall naturally out of foreign keys on height, from each block
to its parents)
ben_vulpes: when i need
to handle reorgs, i'll do so.
ben_vulpes: and i believe
that
the
thing hangs on
to short chains
ben_vulpes: (and only
the ones in
the main chain, atm)
ben_vulpes: i'm leaning on dumpblock
to get
them in chain-order
ben_vulpes: yeah i have
the vast majority of blocks serialized
to disk and am
trimming
them apart.
ben_vulpes is looking for
the log line where asciilifeform made a comment about how
the block structure was
trivially deduceable from
the source, a few days ago after i published
the header serialization snippet
ben_vulpes: i doubt it, as everything so far is little-endian, and only by convention reversed by early block explorers
to show
the zeros first or who
the fuck knows i've never found a sensible explanation for reversing block hashes (and only block hashes!)
☟︎ ben_vulpes: so now
to
the source and hexdump
to see how individual
transactions are structured
ben_vulpes: but
the strange
thing is
that i read
the sequence number where i expected
to see
the index
ben_vulpes: obvs my
thinger is broken in
that
the hash is wrong, but amusingly i get
the sequence number as
tx in index
mircea_popescu: "dad, i want
to put
the cat's head on
the dog's body, can i borrow your hair
trimmer ?" "that won't work, it's
too small and it will get clogged in blood. here's
the chainsaw."
mircea_popescu: gotta be a lot more nurturing in
the way of rope supplies, by your own
theory ?
scriba: Logged on 2016-09-19: [17:19:45] <mircea_popescu> give kid as much rope
to hang self with as kid can carry.
mircea_popescu: the ill gotta be curable at least on occasion, or else you're stuck explaining
this place.
mircea_popescu: but it's also structurally broken ; because it fails
to neatly reduce uncrowned king of universe via reduction
to absurd. which is
the measure of persuasion, apud socrates.
mircea_popescu: anyway. most of my concern is
that your (plural) heartfelt advice is remarkably unpersuasive. gotta wonder wtf devil is at work here.
mircea_popescu: (i'm not even proposing it isn't, just wanna see what
the showing would look like.)
mircea_popescu: trinque yah, but now it's in a much better formulation. at least
to my
taste.
trinque: I will deny he "knows" masamune on
the same grounds
trinque: mircea_popescu: he is *not* replacing
the underlying system and
thus cannot avoid developing comprehensive knowledge of how *it* works before plonking whatever atop it and calling it something
mircea_popescu: trinque his objection is
to
the notion of "learn" and "knowledge" involved in
that statement
though, you might've noticed. and i don't really see it's altogether weak.
trinque: doesn't bother me one bit, only demonstrated
that "I don't know how
to network
these
two machines" and got
thunked
to... learn
that.
mircea_popescu: no, it's not just you, it's me
too. but he didn't ask us!
mircea_popescu: give kid as much rope
to hang self with as kid can carry.
mircea_popescu: it ~is~ as if all you wanna
talk about is why he shouldn't do what he figures he wants
to do.
this isn't very bright, is it ?
mircea_popescu: "dad i wanna get married" "honeybunch, you'll get old and your
tits will sag and it'll suck. don't get married, it's a dead end"
mircea_popescu: but anyway, leaving
the discussion aside for a moment
to focus on
the meta discussion : am i
the only one who's a little irked by
the fact
that kid wants
to do x, gets list of instructions
to not do x ? what is
this,
the nuclear family, elementary unit of
the state ?
mircea_popescu: really, making a lisp
that works for serious applications in
this sense is outside his pay grade