asciilifeform 'enjoying' the wunderful world of 'write one-writer-many-readers mutex from thin air'
asciilifeform: ( cpp14 included one, but 11 -- not. and sane gcc is perma-fossilized on 11... )
asciilifeform thinks 'maybe cribbable from 14's stdlib?'
asciilifeform: prolly not portably tho, likely soup of x64 asm.
awt[asciilifeform]: http://logs.bitdash.io/pest/2022-12-25#1019414 << If you want a patch I can produce it.
bitbot[asciilifeform]: Logged on 2022-12-25 22:07:13 asciilifeform[jonsykkel|awt|deedbot]: phf: atm even that 'send erryone a % command to run' lolbug still standing
asciilifeform: awt: would be great, if you've the cycles, ty
phf[asciilifeform]: http://logs.nosuchlabs.com/log/pest/2022-12-26#1019498 << `he now has two problems`
dulapbot: Logged on 2022-12-26 01:19:39 asciilifeform: thinks 'maybe cribbable from 14's stdlib?'
asciilifeform: phf: went and rtfm'd, and is even worse than at 1st thought ( found wai committee did not include in -11. r-w locks apparently weigh enuff so as to make spinlocking look good in comparison..)
asciilifeform for many many yrs dodged the draft of having to giveashit about locks, semaphores, etc. and nao having to reload into head
asciilifeform: orthogonally -- asciilifeform suspects that his algo from earlier is actually braindamaged. tick() oughtn't be called at 1hz by a thread, oughta be triggered intelligently on a write if the latter happened 1s <= t <= maxticks, per timer, after last write;
asciilifeform: and then purge by tick() floor(t) times.
asciilifeform: ( on read ditto )
asciilifeform: still not gives parallelistic lookup tho. for that, sumthing like rw lock not escapable afaik.
phf[asciilifeform]: tick is essentially gc phase, if you do `smart trigger` you end up with hidden and random `gc phase`. milage may vary
asciilifeform: phf: also tru
asciilifeform leaning in direction of mechanical simplicity, 'optimize laters'