asciilifeform: ipnoje? they have a split screen thing.
asciilifeform: phf: if you're using mine, you will need to fix the sshd config to permit root login ( see the TomServo thread ). i really oughta post an updated image.
asciilifeform: phf: are you setting up a tabletop rockchip ?
asciilifeform: in other lulz, crapple proclaims nomoar support for opengl on their 'os'
asciilifeform: relatedly, mircea_popescu-was-right(tm)(r), asciilifeform's blog nao has a broken stump instead of working commentron, because google's captcha crapola finally gave up the ghost, and nao i gotta figure out how to port mircea_popescu's routine...
asciilifeform: i dun recall if i mentioned : as of coupla wks ago, no moar 'emerge portage' on asciilifeform's-ban-masks gentoo boxen : it now ~demands~ gpg2 turd.☟︎
asciilifeform: ( no expenditure items yet on this proj )
asciilifeform: a rockchip cluster under snsa banner , for pizarro, also under construction ( currently still in component search phase ) , construction also planned for this mo☟︎
asciilifeform: ( tho i am certainly hoping to begin selling pre-cured c101pa boxen in ~this~ mo )
asciilifeform: the 'web dev' nonsense where 'same code written every 6mo' is symptom of a poorly-defined problem to begin with; sorta like wound sutures that reopen
asciilifeform: again in some subfields there is a famine of solid code, in others -- not ( the payware c compiler people, microshit aside, mostly have starved, afaik )
asciilifeform: some activities have this problem, others not ( restauranteurs do not have to contend with a 'i dun need to eat, i already ate in 2011' )
asciilifeform: 'i sold indestructible, perfect $tool to all $n possible buyers, nao what do i eat'
asciilifeform: ( tho, -- iirc it was orlol -- had a pertinent piece, on the unremovable disincentive for market players against selling durable, quality items that never need replacement or 'upgrade' etc )
asciilifeform: junkyard wars (e.g. trb, mp-wp) where one is stuck welding a tank from 5 zaporozhets and 3 lada carcasses, because that's what there is to work with, inevitably are heavyweight☟︎
asciilifeform: ( funnily enuff, asciilifeform has ~own~ 'mp-wp' , made from ~same vintage of wp, that does admittedly only half of what mircea_popescu's does ( i dun have the spam filtration ) )
asciilifeform: ( as described by naggum & elsewhere, in the 'bathtub' piece )
asciilifeform: nao, if phf can make his vtron handle 'arbitrary' (say, up to avail. ram) masses, and without losing anything, moar power to him. but in practice something is usually sacrificed, in the name of speed/efficiency, is the worrisome bit.
asciilifeform: fast comp -- breeds bloat, apparently.
asciilifeform: but history of pc , suggests grimly that they cannot
asciilifeform: all else being equal, tool that laughs at arbitrary masses of payload is theoretically better ( if folx can be trusted to 'keep it in their pants' and not to use it as excuse for bloat )
asciilifeform: when approaching this mass, really oughta start cutting into orthogonal subsystems. and if you can't cut, the design is broken.
asciilifeform disagrees that any proggy's src has any business whatsoever weighing 1MB, much less 9MB
asciilifeform: phf: trb is, what, 800kB, and already imho 'weighs' ~10+ yrs worth of study to fully grasp
asciilifeform: ( mine's in pieces, presently, i am soldering a firmware refill umbilical to it )
asciilifeform: ... this also means that c101pa ( as soon as i have a ready to roll firmware for it ) will be an out-of-the-box gnat dev station.
asciilifeform: diana_coman: all you gotta do is build, then put the tarball on your rc box, extract, and add to .bashrc e.g. PATH="/home/stas/gnat/aarch64-linux-musl-native/bin/:$PATH"; export PATH and you're up & running.
asciilifeform: diana_coman : musltronic gnat on arm64 confirmed to work !
asciilifeform: well recall, this is soft-innerloopmul variant.
asciilifeform: aaaaand hammingweight-independent constanttimeism on arm64 confirmed.
asciilifeform: on ch10 ffa: exponentiator test from end of ch.7 runs in 9.2s on dulap, and 28.4s on the test rk3328-roc-cc machine, producing correct output.