asciilifeform: this being said, still gotta come up with bigendian box to test with at some point ( iirc trinque had one somewhere, but afaik no gnat for it yet )

asciilifeform: observe that 'address bytes only via word slides' worx even if you shat on one endianism and ate on other.

asciilifeform: ( given as you never operate on anyffing that came down the wire without passing checksum )

asciilifeform: mircea_popescu: the beauty is that there's no actual need to standardize one.

asciilifeform: mircea_popescu: how did erdos put it, 'first you forget to unzip after piss, then to unzip ~before~, and only ~then~ your own theorems'

asciilifeform: and sometimes lost entirely. but you prolly knew this.

asciilifeform: diana_coman: keep in mind that udp packets can get arbitrarily reordered in flight. ☟︎

asciilifeform: BingoBoingo: got it, loox same as last time.

asciilifeform: nao, you'll prolly want to queue the incomings, but imho thats outside the scope of this thread

asciilifeform: it's how i did the cmdline thing ( earlier linked )

asciilifeform: ( you let it emit precisely e.g. 512byte into a preallocated buffer that never moves ; the ip it came from, into a 32bit , and that's it )

asciilifeform: imho there's no particular shame in wrapping an os call that doesn't have possibility of overflows

asciilifeform: diana_coman: then you may want to wrap unix recv() as i described ( can use my mmap example )

asciilifeform: but for certain things 'stream' is actually the correct abstraction (e.g. the variable-length tx in classical btc block)

asciilifeform: they baked this sad into file i/o also ( hence how it shows up in 'horsecocks' )

asciilifeform: it presumes that you want unbounded bags of octets

asciilifeform: sorta like the idjicy of gnat's builtin unix cmdline handling

asciilifeform: supposedly will arrive by end of next wk ( in practice, whoknows, iirc the last shot was 1 week of postage and 5 of orc customs )

asciilifeform: BingoBoingo, mod6 , ben_vulpes : ordered ! 308.28 orcbux ( 62.18 of'em orc fee, 37.21 -- postage, the rest -- 8 drives and pack of 20 adhesive hedgehogs ) ☟︎☟︎

asciilifeform: ( imho the tradition where 'just about errybody's little endian, but we spend cpu and flip bytes erry time we send something because in 1989 somebody had a pdp' is asinine ) ☟︎

asciilifeform: to round out thread, http://btcbase.org/log/2018-09-14#1850330 << in my orig udp attempt, i dispensed with the traditional 'gotta be in network byte order (tm)' doctrine, in favour of 'if packet doesn't pass muster, THEN flip the endianism and try again 1ce' ☝︎☟︎

asciilifeform: ( tho will add, ada Strings actually work with arbitrary garbage, given as they do not use the idiot c 'null termination' paradigm )

asciilifeform: ( tho naturally you won't use ada String for arbitrary octetolade. but same principle. )

asciilifeform: http://btcbase.org/log/2018-09-14#1850327 >> they are, see e.g. http://btcbase.org/patches/ffa_ch4_ffacalc#L43 example ☝︎

asciilifeform: http://btcbase.org/log/2018-09-14#1850298 << i'll add to this, that i've been to hobo encampments and found them to be... ~substantially~ cleaner and better organized than the 'open sores' atrocities. i suspect that the shitgnomatic level of horror is only found in dilapidated third world psych wards where inmates live in own shit. ☝︎

asciilifeform: but nao that i think about it, iirc mircea_popescu maintains currently a mswin eulora client, so you might be stuck with gnat's.

asciilifeform: ( the appeal of 'own gnatsockets' is similar to that of what i did with mmap : adacore's implementation is quite obese on acct of handling winblowz and other broken os )

asciilifeform: then again if you write 'own gnatsockets' you would have to do what i did for 'horsecocks' mmap system and actually handle all possible unix eggogs (granted there are not many of these for udp)

asciilifeform: so in this model you'd have 2 routines, 1 which takes a port, listens on it, and comes back 'someday' with a 512byte buffer fulla packet and the originator ip ; and another that eats a 512byte buffer and a dest ip, and returns immediately.

asciilifeform: tcp is very difficult to sanely work with without a stream abstraction, but udp -- trivial.

asciilifeform: diana_coman: incidentally, i was initially quite unsure re the wisdom of using the streams. but they are part of the standard and do appear to work as described. however for your application it isn't even clear to me that you need'em , could simply call out to unix's socket routine and get octet array of fixed length (e.g. 512byte) to work on.

asciilifeform: ( i had'em in the 1st drafts of ffa, then abolished , they interfere with proper compartmentalization )

asciilifeform: the only lang feature that gets in the way of this partitioning is 'generics', but iirc diana_coman did not use these anywhere ☟︎

asciilifeform: diana_coman: and incidentally ( as you prolly already discovered ) you can keep the restrictive flags, on the subset of the coad that doesn't use tasking. ( and this applies in general )

asciilifeform: !Q later tell BingoBoingo http://p.bvulpes.com/pastes/UJFQW/?raw=true

asciilifeform: the only pc-side tool in the setup is (patched) flashrom

asciilifeform: phf: fwiw i do not yet have a working heathen toolchain ( other than what shipped on the box ) either

asciilifeform: phf: this was on the off chance that you had a working replacement rom image. ( if you don't yet -- thing won't do you much good atm )

asciilifeform: http://btcbase.org/log/2018-09-14#1850300 << this is pretty neat. ☝︎

asciilifeform: http://btcbase.org/log/2017-03-10#1624240 << orig thread, re subj, for longterm ref. ☝︎

asciilifeform has reached the mircea_popescu level of amnesia where 'hmm did i make this and it is on hdd somewhere or did i only want to' ☟︎

asciilifeform: hmm, loox like i ~did~ glue'em, but the glued model definitely not ready for primetime.

asciilifeform: this is not directly relevant to diana_coman's application tho

asciilifeform: i then wrote the mmap thing, with aim to bake the 'reads and stores blox , then O(1) tx retrieval' demo, but never got a chance to glue the two together

asciilifeform: ( what it was, was simply a btc block and tx reader, i fed an entire noad's 'dumpblock' chain into it, plus a buncha randomly bitflipped mutilations, behaved correctly )

asciilifeform: diana_coman: re streams, you may find the 'buf_streams' item in horse.tar.gz interesting

asciilifeform: diana_coman: currently i am lacking a bigendian iron (or a gnat for such) so never got a chance to truly and properly test endianism conversion (as for ffa, it is endian-insensitive, but this is simple where there is no networking)

asciilifeform: and without tea

asciilifeform: admittedly asciilifeform just woke up and only gave it 1 look thus far

asciilifeform: i'ma have to repost, 1s

asciilifeform: diana_coman: did you ever see the 'nqb' item asciilifeform posted in '17 ? had various imho good examples of operations on streams

asciilifeform: i also had a variant that used c imports

asciilifeform: i've a half-written 'g' (if anyone recalls what that was) , used adasockets

asciilifeform: ( and i didn't get a chance to finish the whole thing, was right when i picked up ffa )

asciilifeform: http://btcbase.org/log/2018-09-14#1850323 << i have a proggy, but it dun seem like i ever published the item, so diana_coman will have to count as 1st ☝︎