44200+ entries in 0.027s
a111: Logged on 2019-02-20 15:02 mircea_popescu:
trinque did you ever see
the guy irl lately ? say
this year or something ?
Mocky: but i don't have asciilifeform's mental model of a much better lang with which
to compare. I have c c++ python javascript
that I also get paid
to work with occasionally, which ~entirely punt on
threading
a111: Logged on 2017-03-30 14:50 asciilifeform:
trinque: erlang wasn't simply about 'uptime', or even 'no pointer arithmetic', it also was
the only case i know of where process migration actually worked
a111: Logged on 2017-03-30 14:43 asciilifeform: anyway erlang is imho only worth discussing as part of a larger pattern -- industry after industry independently discovered
that c -- and its entire approach
to logic -- is poison
a111: Logged on 2017-03-30 14:38 asciilifeform: mircea_popescu: i will also nitpick : 'erlang' does not belong in
the list, it was a 1980s product
that worked quite well in its industrial niche (large
telco switches) but was later stolen and used as a
totem by
the folx from yesterday's
thread (
http://btcbase.org/log/2017-03-29#1633873 )
a111: Logged on 2015-06-22 21:22 mircea_popescu: ascii_field
this goes right into our discussion about how
the only reason lisp is ok where c is shot is
that
the hordes haven't sat on lisp but on c, would have been
the other way around if
they had sat
the other way.
Mocky: erlang I did look at briefly, but not
the concurrency, seemed
to me like a puzzle lang
Mocky: fortran
threading I never looked at, but will. pretty sure
threw away all my punch cards
tho
Mocky: iirc
that was after backus
turned
to shit in committee
Mocky: serious question asciilifeform, where do I look
to see better
threading
that java. I honestly dont' know
Mocky: also
threads in infinite loops can always be stopped on every platform
Mocky: i wouldn't argue
that explicit
threading is good. but instead of concurrent i++, consider concurrent assignment
to heap references, even without locking always have a valid reference, never a garbage pointer, no matter how many
threads or cores
Mocky: i've seen dozens of
those. look equally shit on every platform
Mocky: I'm not defending java, I'm stating
that in addition
to pile of shit,
theres a coherent memory model,
thread model
that is not agony
to work with.
Mocky: yeah, early java lists were shit in
that way but no more. now shit in diff ways
Mocky: so not optimal, maybe even laughable, but yet has a model and as spec
that is not self contradictory
Mocky: laugh if you will, yet still can't do it even *with* locking in standard c++, as far as I can
tell
Mocky: and yes can increment global and local counters in
threads and have it add up, but requires locking or cas
mircea_popescu: Mocky just
trying
to evaluate if
this
thing worth digging up
Mocky: i'm not saying
they solved
threading once and for all, or
that eliminated deadlock. any system
that allows you
to acquire locks in different order can be deadlocked. i'm just saying
that
there are concurrent primitives
that can be understood and which have guaranttes
that hold
mircea_popescu: but
the ensemble will not move at
the speed of a single core will it ?
☟︎ Mocky: can do multithreaded i++ and never get garbage, if you mark it volatile forces read and write
to main memory
mircea_popescu: Mocky so if i declare a procedure which increments a global and a local counter, and
then call it in a loop from multiple
threads, at
the end
the globa lflag will contain
the sum of
the local flags ?
Mocky: there are explicit and implicit semantics, but if you read
the spec, you know what you'll get. multiple
threads setting a value on
the same variable never create garbage
Mocky: I mostly agree.
They did manage a coherent
threading model / memory model which
turns out
to be
the
thing I miss when I'm writing in something else
a111: Logged on 2019-02-20 15:50 asciilifeform: really oughta have ~1~ compiler, and for a sane arch; and if yet cannot afford
to actually siliconize
the sane arch,
then
the sad iron oughta be booting straight in bios
to a compact (asmistic) emulator of same
Mocky: I was hopeful back in
the day at java chip and java os, nothing worthwhile surfaced
a111: Logged on 2019-02-20 16:08 asciilifeform: various folx attempted something roughly like
this, but succumbed
to own idiocy -- e.g.
the java people, who 1) wrapped
their 'vm' around
the nonsensical shape of
their
tardlang 2) stopped short of 'and no
this doesn't run on your
tard os, is IS
the os, fuckyou' ;
BingoBoingo: Dead cop further down page
than dead graffiti communist punk
BingoBoingo: A bit deeper "Uruguayos no somos Latinos, somos Europeos": "Mi amiga
tuvo un hijo en EEUU hace un
tiempito y cuando lo van a anotar, dentro de las cosas que ponen en la partida es raza (WTF EEUU??!), la administrativa no quería ponerle "caucásico" así que le puso Latino white o una onda así."
BingoBoingo: From
the mines: "Viví 6 meses en Madrid y varias veces se me rieron en la cara (la mayoría empleados públicos, y una vez en un banco) por mi manera de hablar, pero mi peor experiencia fue en entrevistas de
trabajo donde me dijeron que el problema era que no hablaba español." - Some Uruguayo
hanbot: right right, i mixed 'em up. and yeah, i'm planning on grabbing phf's keccak v.py in step 3, if only because i've seen diana_coman's pop up in cuntoo
tests so i'd like
to
test
the ver less
traveled.
mircea_popescu: hanbot write down
the whole story as you go, but basically yes.
hanbot: where "knows
to v" = "knows no v". heh.
hanbot: in bootstrapping adventures, it looks like
the flow for a machine
that knows
to v
to get vtools going is like so: grab some ancient v, ie mod6's v.py, use it
to press phf's vtools vpatch,
then eat phf's v.py "updated for vtools",
then press vtools
to keccak head. anyone feel like spotting
this for me?
mircea_popescu: conquering china, also expensive. who knows, maybe it can be massaged
to pay for itself.
mircea_popescu: kinda makes it a particularly delicious spot
to rape
the femstate.
a111: Logged on 2016-01-19 21:55 ascii_butugychag:
the whole point of engineering is
to cut apart
the happenstance-linked crud from what you actually want.
a111: Logged on 2016-06-14 14:12 asciilifeform: did we ever do a screw
threads
thread ?
a111: Logged on 2017-04-09 23:14 mircea_popescu: "who do you
think will win, mr 13,
the
team with
two guys and monkeys or
the
team with
three guys and monkeys ?"
mircea_popescu: ie, whenever you encounter "competitions" of malformed nonsense, you're exactly in
the position of
the ~spectator~
to special olympics.
mircea_popescu: i suppose
the pov where design is ~constrained~ activity, and
the concept of "design competition" must strictly describe a situation where machine a with 64 registers of which one is mask and machine b with 64 registers + ONE separate mask register compete.
mircea_popescu: as a general design principle! you may NOT have "multiple" on a machine. it's 0, 1, buswidth-count.
THAT IS IT.
mircea_popescu: let me guess,
the same fucking people who made mul without carry.