asciilifeform: http://btcbase.org/log/2017-12-03#1745763 << my current reading of the shoemaker is that he was some combo of senility and 'tilt' (being under attack from ~all angles by the empire, and unable to do much of anything in the way of genuine push-back )☝︎☟︎
asciilifeform: http://btcbase.org/log/2017-12-03#1745762 << the pizzeria people didn't do anything that, e.g., bush's 'wmd' people didn't do. only diff is that one did it while inca emperor, and demolished a good chunk of whole continent, while other -- did it while penniless hobos, and perforated a roof in an old house.☝︎
asciilifeform: could, certainly, be a case of 'there are only so-many cheapo clip arts'. but also not a bad 'subconscious confession' of muppetry, alternatively.
asciilifeform: on right hand -- 'simon', american children's toy where coloured lights blink and kid tries to repeat the sequence
asciilifeform: http://btcbase.org/log/2017-12-03#1745755 << it was a formatting issue, and he repasted with p.bvulpes -- but the unfortunate bit is that they, too, will vanish, and the comment becomes useless in <1week unless i manually archive'em☝︎
asciilifeform: for certain purposes it makes a permanent difference.
asciilifeform: anyone naturally can cut up into whatever pieces. the issue is how is it structured on asciilifeform's desk.
asciilifeform: ( btw nobody yet asked why ffa is now a standalone-built static lib, rather than as before , a set of sources to be simply dropped into other projects. answer is -- binary audit )
asciilifeform: ( knuthian divider, if anyone forgets, is an exact inversion of egyptian multiplier. )
asciilifeform: and solely knuthian divider for the modulus op. etc.
asciilifeform: ( 'ideal mathematically-sufficient' ffa would probably include solely egyptian multiplier, for example. )
asciilifeform: but there are already plenty of haskellists writing nonsense. asciilifeform wanted an actually usable fits-in-head item.
asciilifeform: and if i were a haskellist and writing with an eye toward ~machine~ proof, rather than fits-in-head, i probably would have written one.
asciilifeform: diana_coman: you do have a point , though, an ffa that is an ideal mathematical description of every relevant bound, would look slightly different from the ffa i actually have, that has to actually run on idjit cmachine iron in something like real time.☟︎☟︎
asciilifeform: so generic -- gone. ( also shrinks the binary, a bit )
asciilifeform: but it turns out that it is not possible to have a standalone static lib containing an uninstantiated generic, in adaworld
asciilifeform: the genericunit allowed a given running instance of ffa to know the actual ffawidth desired; and to have stricter checks on arguments
asciilifeform: the latter required finalization and for FZ to be a struct rather than bare array ( finalization massively bloated the binary; whereas structization massively bloated the src )
asciilifeform: and originally i even had autozeroing for FZ, and zeroing on going-out-of-stackframe.
asciilifeform: readers may recall that originally ffa was a 'generic unit'
asciilifeform: a value of type FZBit_Index cannot be assigned a value greater than machineaddressspace/wordsize without 9,000 other things being unfathomably broken
asciilifeform: i did not do it because it adds runtime overhead ( the bounds get checked , and even in inner loops ) without preventing any kind of problem that can happen in reality
asciilifeform: diana_coman's observation would make sense for narrowing the type of FZBit_Index .
asciilifeform: Indices is the root type for ~all~ possible indices, of anything.
asciilifeform: what does dividing it by word bitness ( Word'Size ) accomplish ?
asciilifeform: theoretically a word count of a number equal to the number of words in machine's memory, is legal
asciilifeform: 1 .. Indices'Last is simply a way to say 'exactly same as Indices, but 0 is forbidden'
asciilifeform: diana_coman: why would you divide ? i dun get it
asciilifeform: whereas most of what folx think of as traditional radio spectrum, goes around corners, terrain, moar omnidirectional flavour
asciilifeform: also depends on wavelength. uhf (10s of GHz) line-of-sight beams behave muchly like laser, you gotta know exactly where it is and sit down into the path through air, no bigger than train tunnel, to listen
asciilifeform: errythingdepends, i'd expect, on the relative 'sizes' of fox vs forest trees.
asciilifeform: the ho chi minh truck search iirc worked better because smaller background.
asciilifeform: mention of 'fox hunt' thread led me to think that the iraq memory were re rf triangulation, rather than search for ir-glowing truck engines from aerial camera
asciilifeform: but. but i can think of one scenario where they are not pointers, explicitly, but can still have'em point to same place. (though i also banned it!) --- if there were an allocator pool that somehow put X and Y, conventional nonpointerolade items, in the same exact place in memory.
asciilifeform: normally you would have to call it like foo(X.all, Y.all) .
asciilifeform: if pointers were permitted -- it could occur in a foo(A, B) when A turns out to point to same thing as B
asciilifeform: mircea_popescu: my current understanding is that , having banned pointers, it can only occur with an explicit call foo(X, X) , visible to naked eye hamfistmistake in src.
asciilifeform: somebody oughta get order-of-lenin for these caught wreckerz.
