asciilifeform: ( what 'genericism' does in ada, is to force the compiler to rebuild a copy of the lib for erry time the exported type is invoked in the caller. at which point may as well tell user to drop the src into his, rather than link the lib, the effect is the same )
asciilifeform: the 1 workaround i've found is to make the lib 100% generic 'all the way down', but it's ugly imho
asciilifeform: ( which is why my mmap thing aint posted yet )
asciilifeform: gnat is entirely ok with finalization in lib so long as the finalized type aint exported☟︎
asciilifeform: bvt: of course it does, trb is 100% static-linked and has plenty
asciilifeform: ( for old c/cpp dogs : 'finalization' is analogous to 'destructor' )
asciilifeform: the standard, btw, does not say any such thing as 'using finalization breaks staticity', it seems to be 100% gnat breakage rather than 'standard forbids'
asciilifeform: current ffa relies on the user/caller to wipe
asciilifeform: ( it's not a wholly useless thing, i dun have it in current ffa but my 1st draft used it to wipe all data on the stack erry time it goes out of scope )
asciilifeform: some language features seem to break the 'encapsulability' , e.g. finalization
asciilifeform: this pattern is not really escapable if yer using the restrictions
asciilifeform: bvt: i structured it so that the moar-restriction-y stuff ends up in static libs that get linked ~into~ the less-restrictiony demolade ( rather than vice-versa ), and so that the lib is statically linkable to begin with
asciilifeform did quite a bit of gnat sores archaeology dig
asciilifeform: bvt: it's structured the way it is, for a reason
asciilifeform: ( who outside of us is linking statically... )
asciilifeform: i also suspect that my use of static linkage exposed buncha implementation strange
asciilifeform: overall pattern is frequently 'aha, you banned this ? goodluck using the standard lib nao, where we used it 9000 times'. saw this with e.g. the secondary stack liquishit.
asciilifeform: really calls for a deep dig into the sores.
asciilifeform: ( how gnat goes about enforcing the various restrictions, is largely black art at this point, the most i can say is that it seems to 'fail deadly', i.e. ultra-conservatively banhammers yer entire proggy if it fails to satisfy various obscure this-and-not-that conditions )
asciilifeform: mircea_popescu: didn't say it was 'same thing', yes it comes with the grandfather of all shitstacks (20+GB of liquishit) . it's for simulating designs, not for deploying ( why wouldja deploy anyffing on such a thing, it makes cluster of pc look vehehery cheap and 'opensores'y by comparison )
asciilifeform: 'your princess is in another castle'(tm)(r) eh☟︎
asciilifeform: Mocky: sounds like exactly what mircea_popescu was speaking of. see if allah wills that he shows up to his desk..☟︎
asciilifeform: but they're actually in my parts houses' catalogues.
asciilifeform: afaik they're ~only~ used for asic dev, hence the 'we charge weight in diamonds'
asciilifeform: they offer a half-mil. LUT thing. (can fit, e.g., orig 'pentium', say.)
asciilifeform: ( naturally it's the big-fpga monopolist, xilinx . but it's there. )
asciilifeform: apropos of upstack -- last wk asciilifeform did the ~yearly dig re 'does anyone actually sell fpga big enuff to demo 8192b-arithmetizer inside, fully unrolled' and turns out that yes (as of 6mo ago)
asciilifeform: mircea_popescu: i get pingbacks from these nonstop. tho more often they're shannonized 'miner' sites
asciilifeform: mircea_popescu: depending how you cut it, varies . e.g. my 'eizo' has a 'byteness' of 14 bits ( per channel )
asciilifeform: mircea_popescu: for so long as you can fit in $process -- yes entirely same
asciilifeform: my orig observation was that this is a less stupid use of die space than e.g tlb cache and the other liquishit x86 is packed with 'so win10 faster'
asciilifeform watched in horror as the cabbage ate teh wolf, last time he played w+g+c
asciilifeform: https://archive.is/wniiv << meanwhile in vintage lulz. d00d built wankatron with spoon in arse , plugged into mains , with neon lamp to reduce 220v , or so he thought.. it shorted.
asciilifeform: optimally , by the time the scissors actually come out, there'll be concretes in the l0gz, of the form of 'hands off my $scsicard, motherfuckers'
asciilifeform: mircea_popescu: would like to get sense of what is thought 'obscura' tho, my obscura may or may not be trinque's obscura, etc.
asciilifeform: ditto s390, though if somebody insists on ibm dinosaur in his tmsr workflow i'ma be curious re wtf,wai
asciilifeform: anybody who finds that he cannot live without one, is welcome to backport the kernelade for it with ~own~ hands, and present signed vpatch.
asciilifeform: i dun have e.g. 'hauppauge tv card' and dun intend to dig in junkyard for one just to test driver.
asciilifeform: BingoBoingo: i seem to recall you had one ? at some point
asciilifeform: ( 68k has, what, 24MB ? max addr space. to let 'modern' kernel eat 12-20 of these, is imho pretty 'rich' )
asciilifeform: who the fuck maintains that branch, i want to know, and on what !
asciilifeform: for that matter, errybody luvvs m68k, but where might one actually get one with enuff ram to fit the current kernel.
asciilifeform: ( does anybody even ~have~ a s390, or where to steal one ?? )
asciilifeform: at some point i'ma get around to trimming the kernel. how many of these do we actually want ?☟︎
asciilifeform: i've built, fwiw, for mips ( 32 an' 64 -wide ), for arm (ditto), even for 'microblaze' ( spoiler: not an actual iron, but xilinx's oddball 'demo cpu', exists nowhere else but 'virtex' demo boards, of which i have an embarrassingly tall pile ) -- but i ain't ever built for s390, or 68k ...
asciilifeform had the mysql manual open , since BingoBoingo 1st mentioned subj, and somewhat to my surprise found that the thing dun seem to have a max mem knob !
asciilifeform: ( i still see it as 'railroad gauge' , gonna publish 80cols till the day i die )
