asciilifeform: mircea_popescu: what we ~can~ do is to proclaim that nodes having finite tx capacity will order by value-per-bt and order connection priority likewise by avg. value/bt of incoming tx, and so on recursively.
asciilifeform: the other side of this medal is that tx is simply too cheap today.
asciilifeform: approx. the same. result will look quite like, e.g., reiserfs.
asciilifeform: the correct solution to this is more or less equivalent to writing a filesystem.
asciilifeform: then you do the next arithmetically-adjacent tx, ad infinitum, until you have N contiguous bts free
asciilifeform: but what you ~can~ do is 1) pick a tx which must move 2) declare it dead temporarily, copy to a cache 3) introduce it as if it were a new incoming tx.
asciilifeform: naturally, you can't actually move a cpp obj because of pointer spaghetti.
asciilifeform: i was about to describe the only possible way to make it work.
asciilifeform: (if you can't relocate, YOU CAN'T DEFRAG)
asciilifeform is plagued by the suspicion that we are trying to remove a 20kg metastatic tumour using.. acupuncture
asciilifeform: 'The next idea was,that heap is sprinkled with permanently allocated blocks data causing inability for malloc to release the memory back to the OS. I went to try and allocate these separately... and boy does that rabbit hole go deep...' << ty jurov, i was actually in the middle of doing this, and now i don't have to, l0l !!
asciilifeform: jurov: i've been bashing my head against the thing pretty actively, on and off, for >1yr now. so you're doing pretty well.
asciilifeform: jurov: pretty interesting, i'll say more after i actually read the thing
asciilifeform: mircea_popescu: 'Тяжело в деревне без нагана. И с наганом - тоже тяжело.' (tm) (r)
asciilifeform: mircea_popescu: complicated. i have a human (non-gmail) mailbox that can't use mircea_popescu's algorithm (has to be able to receive from strangers) and it costs me perhaps 20min/day of manual spam filtration
