341600+ entries in 0.215s
mod6: so ben_vulpes, understand
that
there is like a jumble of different
things he submitted, and if
the wrong ones are incorporated, it might puke.
mod6: i kinda have
to
turn a deaf ear
to a lot of
this stuff until
the new month starts, and even
then, i gotta keep my eye on
the prize - dispite any other problems on
the periphery.
ben_vulpes: so if i press
those individually, in order,
they work. but if i press
the last patch, i believe
that it fails
to find its antecedents.
mod6: lets not worry about
that for now.
that
thing is in its infancy.
mod6: just had
to stick his stuff into patches and .seals. but in all honesty, im not even sure if
the patches & seals I'm using are what is indended at
this point.
mod6: i was able
to press all of
that out just fine.
mod6: yeah, no clue what you did
there.
mod6: i might need a reminder. i did go
through a bunch of your comments
though.
ben_vulpes: did you get a chance
to repro
the issue i ran into with asciilifeform's shiva patches?
mod6: im gonna
test
this revised patch, if works, send it,
then get
that SoBA in ship-shape.
ben_vulpes: entirely unrelatedly, can anyone corroborate
the story of
the russians who burned a shitton of usd on camera and got
the
then-fed
to credit some domestic account or other?
ben_vulpes: how many rickety fucking buckets of others' design must a man endure before he gets
the fuck out and walks on his own feet is
the question i'm endeavoring
to resolve for myself
adlai: how many roads must a man walk down before he stops
taking
the less-trod fork
adlai: in
the same sense as "eleven rabbis,
twelve opinions"
adlai wonders at
the boundedness of CL vtrons... probably nonexistent
adlai: phf:
this web-facing
thing is part of your CL vtron?
assbot: Logged on 31-01-2016 19:45:44; phf: clasp is not "compiler research", is designed
to scratch itch of a single person, for doing computational chemistry
mod6: I'm going
to build with
this and live
test it before I submit
to
the mailing list.
adlai: well patches, not seals.
the latter aren't particularly stimulating reading material per se
mircea_popescu: nevertheless,
the fact
that bitcoin-os is
the unavoidable end of
this entire process requires
things be
thought out with a view
to
that fact.
mircea_popescu: but in any case, compiling bitcoin in bitcoin is not either a goal or liable
to be implemented before actually having a bitcoin-fs say.
mircea_popescu: i
think maybe you've formed a partial idea of what's being discussed
through
the age old process of having read a partial set of what's been said.
adlai doesn't want a CD because 4/5ths of his hardware doesn't know what CDs are... but
this is back
to bedrock issues
adlai was "it's"ing about "doing exactly
that", not about "BOOT CD"
mircea_popescu: looky : currently we have a script
that pulls dependencies, checks
them, and builds.
adlai: mircea_popescu: in
these early days, when Bitcoin does not yet fully Exist, it seems as
though everything is an option.
the full shiva vision, as i understand it from recent logs, opens
the door
to doing exactly
that, if some energetic idiot decides it's a good idea
mircea_popescu: adlai
there's no option
to embedding a compiler in an os.
phf: clasp is not "compiler research", is designed
to scratch itch of a single person, for doing computational chemistry
☟︎ adlai: again, not
trying
to perscribe what's right; obviously asciilifeform understands what he's
trying
to do infinitely better
than i can; but... i
think
the line is somewhere between embedding an interpreter for a subset of r5rs, and embedding a ~compiler~
mircea_popescu: this log is moving so fast i'm actually getting contiguity issues
trying
to follow.
adlai: ergo why
the clasp horse is also unnecessary
to continue flogging.
adlai: lemme put it another way... somewhere in
the ANSI spec are probably one,
two, or even
three ~hundred~ symbols which
the weaponized bitcoind-os does not need in its guts. define-setf-expander? define-method-combination? where do you draw
the line?
mircea_popescu: adlai no, shiva isn't supposed
to be "an optimal point". it is supposed
to be a CORRECT replacement of a currently broken implementation.
assbot: Logged on 31-01-2016 15:28:44; mircea_popescu: asciilifeform because
thiel got money and no self respect.
mircea_popescu: phf but
these adnotations are necessarily not
translateable.
adlai: rather
than some continually-shifting goalpost
that asymptotically approaches whatever sentience seems least broken
phf: asciilifeform: while
the goals are different,
the mechanism is
the same. you annotate classes, methods and get stable refs in your lisp instance, which shares heap with
the running c++ process.
adlai: asciilifeform: isn't shiva supposed
to be an optimal point BETWEEN sentient-AGI-bitcoinOS and jsonrpc on port 8332?
mircea_popescu: well it'd pretty damned well be able
to, else how is it an os
adlai: my understanding is
that nobody here wants
to make bitcoind,
the minimal concrete-sealed weaponized Bitcoin-that-Exists, able
to compile a new version of itself
adlai: while you may want missile guidance on board
the missile, you might not want
the chip fab on board as well
phf: asciilifeform:
that's what clasp does
phf: adlai: i don't see
the difference. both require you
to annotate your classes/method/data somehow. clasp requires less annotation with larger impact
then
tinyscheme, e.g. it knows about c++ and can do some advanced ffi-ing
mircea_popescu: point remains, resource needs don't scale well in
this
type of problem.
adlai: phf: also, afaict, clasp is designed for manipulating existing C++ libraries with your own
tendrils, rather
than
tendriling an existing
turdatron.
tinyscheme is designed for
the latter.
mircea_popescu: well so
then. i don't want 10x as many people
that can jump 1m. i want people who can jump 10m now.
mircea_popescu: as
the linecount you're digesting increases you don't need more people. you need smarter ones. it's not like if you need one brick carried you hire one lout and ifg you need 10000 you hire 10000. it's like if you want 1kg carried you can use a dog, and if you need 10
tons you gotta learn
to speak whale now.
adlai: modern CL (as opposed
to
the ANSI doorstop) is a
tool
that evolved from a standard. r5rs is a standard.
tinyscheme is a
tool. etc etc... vita brevis, but
there are at least a couple usable artes
phf: adlai: clasp is a good fit for
tendriling
trb, if i had more
time, i'd use it in a similar way
to shiva. but
then i'm already building
trb on llvm/clang, etc. it is anti-ba
though because of pedigree and ownership problem. can use it, if you have resources
to read
through 5 or so 100k loc projects
mircea_popescu: physics is a well known whore. behaves
the exact same way.
mircea_popescu: ben_vulpes you know what "not making
the cut" is wqith a woman ? even if you're a saudi prince and your daddy buys her for you ; even if you're an afghan monkey and she deeply believes she has
to.
there's a point where you make
the cut or don't, and
that's what it is.
adlai: clasp is an answer
to a specific problem which nobody here has... it's equally anti-pythagoras
phf: clasp is very much anti-ba,
there's no question about it, pulls a lot of projects
together, of questionable pedigree
mircea_popescu: <adlai> if "standards" is plural, it's already noncompliant <<< for
the record,
the reason
the enemy uses
the plural
there isn't
that
the enemy is
this stupid, but
that it believes in patchwork. ie, standards : one for
this, one for
that,
together
they make up a complete
thing.
ben_vulpes: adlai: i don't know shit, frere, i just drag in dead birds for
the adults
to dispose of.