asciilifeform: cruddy connections , afaik, is the only recurrent problem
asciilifeform: lol! 'why do we live in the train tunnel, we could afford flat' 'they haven't kicked us yet...'
asciilifeform: quasi-relatedly, asciilifeform found himself wondering, why we're still on fleanode... between pizarro, mircea_popescu's racked boxes ( idk where, but presumably good ones ? ) , asciilifeform's own fiber, trinque , etc. could have a very solid irc net, potentially☟︎
asciilifeform: probably i ought to have pointed this out moar clearly.
asciilifeform: mod6: there's not an urgent and dire need to test the linked item ( though fwiw it stood up to exhaustive test on asciilifeform's side ). i linked it in response to http://btcbase.org/log/2018-05-01#1806923 thread strictly.☝︎
asciilifeform: and the 'tldr' of it is: if you download the patches/sigs in the example, and press to any particular one, you get a file 'trb', which, when put through included proggy 'txt2dir' results in a bitwise-correct (i.e. bit-identical to classical tree that we have for trb) press.
asciilifeform: mod6: the linked item was written in as illustration of 1 possible solution to problem posed by mircea_popescu and trinque , in re spuriously-independent patches creating misleading graph
asciilifeform: mod6: generally it's good practice to send in the eggogs some time near the time you actually did the test...
asciilifeform: ( if you grab the attachments of the ml post, and follow the instructions, you get bitwise-identical trb on every step of the ladder , vs the respective trb at that step in the orig tree . )
asciilifeform: mod6: considering that the example i gave ~is~ the totality of the trb tree, i'm a bit puzzled, what was it you tried ?
asciilifeform: funnily enuff this is not how orig gosplan worked. ( there was a multitude of -- de-facto independent -- factories , often overlapping in subj matter )
asciilifeform: i.e. to take over the old functions of usg.microshit.
asciilifeform: and in long term to set the stage for 'we finally standardized pc hardware' ( on googlenintendo naturally )
asciilifeform: think, they collaborated on 'killing' gcc ( to be rid of the human mushroom ) and nao ditto for linuxkernel ( to be rid of the not always-cooperative linus )
asciilifeform: ben_vulpes, mircea_popescu : what sort of sv usg.corp would it be if it did not ( like e.g. crapple ) have own 'microkernel' abomination
asciilifeform: seems to contain a number of kindergarten c/asm tutorials by some orc, and little else
asciilifeform: asciilifeform's orig point was precisely that... the clothespins dun do nuffin for asciilifeform . may as well be a hat with feather, or 9000 other things he's entirely insensitive to. and even the skins, i prefer the unedited ones. just-so.
asciilifeform: the one on right hand, has moar skin an' moar photoshop... not 'fair fight'
asciilifeform: see, if one were to ask asciilifeform , even a haphazardly-thrown dart into 1980s su will hit one, e.g. https://archive.li/2VPRj ( tbf does include some older era pics also )
asciilifeform: not seekrit, but unportable; we're unlikely to agree .
asciilifeform: oh hey is it based on that oddball 'nazi' treatise ?
asciilifeform: ( and http://btcbase.org/log/2016-10-31#1560740 , and there was a particular link some time in past yr which i now cannot find, where someone actually went and ~measured~ the reaction time delay of msdos, various winblowz boxen from past 20yr, crapple, bolix, etc etc and found exactly what i'd expect him to find )☝︎☟︎
asciilifeform: in, say , 1950s, they had many fewer engineers (there was not this peculiar compulsion to conscript errything with a pulse into 'engineering' ) and the ones they had, designed autobahn ramps, there was all the employment they could use
asciilifeform: dunno that it even works in this direction; granary with busted walls is not 'to provide employment for 9000 mice', they just show up.
asciilifeform: they stave off the complexicollapse, same way as the fed pushed forward the monetary one
asciilifeform: the other driver, given as the 'nm shrinkage' is ~100% wintel-driven -- 'moar room for microshit bloat'
asciilifeform: ( iirc not long ago there was thread re the world's only 10nm fab, belonging to usg )
asciilifeform: aaha. the smaller the 'nanometrage', the moar incatronic the process.
asciilifeform: trend's been entirely towards 'britneyization', 'economy of scale', 'build 500x larger and costlier production line but make $1 cheaper dies'
asciilifeform: mircea_popescu: in ~factory~ machinery, afaik the trend's been quite the opposite ( small manufacturers of ic existed in 1970s by the thousand, in 1980s -- by the hundred, today -- gone. )☟︎
asciilifeform: ideally you want a 'polaroid'-sized box, i.e. a fungible cartridge that starts out as coupla grams of sandwiched $layers, and ends up after exposure as $ic
asciilifeform: ( and somehow admit the feedstock like film is admitted into camera, i.e. without introducing dust )
asciilifeform: 'clean room' solves itself if you could somehow get 100% of the process to fit in a reasonably small hermetic box..
asciilifeform: and HF is same bitch today as in 1800s.
asciilifeform: mircea_popescu: doubtful, for same reason that electron microscope has not become any cheaper. high vacuum system is a bitch, and exactly same bitch today as in 1930.
asciilifeform: 'polaroid' is not the only possible scheme for ic sanity. could also have something akin to current-day 'dye sublimation' printer.
asciilifeform: not to mention the 'clean room' business, where 1 flake of skin easily is enuff dirt to ruin ~several~ chips
asciilifeform: PeterL: indeed it does. however it also involves sputtering ( in, naturally, high vacuum ) to deposit metal, HF (and other nasties) to etch it away , and various other painfuls
