asciilifeform: ben_vulpes: exactly like every other 'let's tor, let's wank to isidora whorecraft watshername' folx, neh
asciilifeform: mircea_popescu: and yes, it was exactly 'jackpot shirt'
asciilifeform: i've picked up 'old-new stock' before, but this is somethingelse.
asciilifeform: in other lulz, asciilifeform unpacks a small crate of b00kz from ru, and finds that the rus-ro dictionary, shrinkwrapped, but inside : item from 1942 !
asciilifeform: usg.blue quite like the stereotypical idjit gambler, wearing ancient crusty shirt he once jackpotted in
asciilifeform: they've been replaying the exact same program ever since it worked for killing the old shoemaker.
asciilifeform: if you remove the div0 trap from ffa, you end up likewise with a reg full of 1s. it's what knuth division produces when given a 0 divisor.
asciilifeform: it isn't distinguishable from a conceivable legit division.
asciilifeform: to say 'error flag' is not in line with reality of the situation -- the only correct answer to div-by-0 is eggog. and eggog != 'returns a value just like actual non-by-0 division might'
asciilifeform: that's not an error flag -- negative 1 is a legitimate result of dividing, e.g., 100 by -100
asciilifeform: '-1 was value seen in other simple implementations and drops out of simple hardware implementations.'
asciilifeform: 'The lack of a carry bit complicates multiple-precision arithmetic – GMP, MPFR ... RISC-V does not detect or flag most arithmetic errors, including overflow, underflow and divide by zero ... Most popular programming languages do not support checks for integer overflow, partly because most architectures impose a significant runtime penalty to check for overflow on integer arithmetic and partly because modulo arithmetic is sometim
asciilifeform: so it's a first-class wrecking. sorta like what llvm used to be prior to apple crowning it. riscv is waiting for its apple.
asciilifeform: they also managed to waste the time of compiler back-end people, linux kernel folx, etc. with it.
asciilifeform: but nobody afaik sells a siliconized hard implementation of it. and i dun expect that anyone will. because wtf.
asciilifeform: moar concretely they go around propagandizing that if you build a fpga board you gotta use it, because 'mips is patented' ( in what country ? even in usa patent is 25y ) or 'risc-v is simplest' ( uh, nope ) and other idiocy.☟︎
asciilifeform: ( and why my 1980s serial boxen still work, etc )
asciilifeform: and why phf is able to execute 40 y.o. macsyma , ditto
asciilifeform: standard compliance (we also had thread) is a massively underappreciated thing. it is why, e.g., diana_coman was able to take the serpent thing and build it, cleanly and without change, 20+yrs after its writing
asciilifeform: it isn't a straight imbecile mapping to the idjit cmachine cpu, as c is.
asciilifeform: but as we discussed in old thread, the single most serious problem of ada, is that it is quite difficult to implement an adatron. esp if you actually go for standard compliance.
asciilifeform: to try to make analogy between world's single most supported electrical standard after 220v mains socket, and intel turdolade, is beyond ludicrous.
asciilifeform: but the logical protocol is classical rs232, 115200/8/1/noparity.
asciilifeform: you likewise need exactly same voltage converter for pogo , and most other small devices.
