awt[asciilifeform]: !!ticker btc usd
signpost[asciilifeform]: I noticed jurov's source viewer is now down, RIP
signpost[asciilifeform]: I think I might've found a bug. the BASTARD block branch of ProcessBlock in main.cpp returns true, but the block hasn't actually been accepted. blocks are also added to inventory regardless of whether they were stored or thrown out in the block handler of ProcessMessage.
signpost[asciilifeform]: I changed the return to false, and made the inventory update conditional on block acceptance, and am seeing whether that does anything
signpost[asciilifeform]: scratching my head at seeing miles upon miles of "BASTARD" in logs sent me down this rabbit hole.
asciilifeform: (where ProcessBlock called.)
asciilifeform: if you return false, the shitblock's hash aint removed from 'askedfor'
asciilifeform: 'not bug -- pheature'(tm)
signpost[asciilifeform]: hm. that part makes sense, but why would I keep it in my inventory if I didn't keep the block?
asciilifeform: 'Open Secure Chip'(tm)(r)(c) etc, but where's e.g. the quartz lid or 4uM process etc
asciilifeform: http://logs.bitdash.io/pest/2023-05-24#1026314 << iirc kaput ~2y ago, which was wai cgra baked new one
bitbot[asciilifeform]: Logged on 2023-05-24 10:46:52 signpost: I noticed jurov's source viewer is now down, RIP
signpost[asciilifeform]: I probably already knew that and forgot
signpost[asciilifeform]: ah wait, I missed a block, sec
signpost[asciilifeform]: I thought "asked for" and "known" are separate concepts, and it looks like bastards are considered known even though they were discarded.
signpost[asciilifeform]: but this is me poking the thing with a stick, not a proposed patch
asciilifeform: signpost: gotta somehow prevent the 'i threw it out but will ask for it again right away just the same' loop, neh
signpost[asciilifeform]: will have to pull more of this shitsoup back into my head to say, could be right
asciilifeform: ( ... or not, but tradeoff b/w ram guzzle & bw guzzle obv )
signpost[asciilifeform]: right now the node's syncing with it, but doesn't mean it can't end up looping
asciilifeform: the unfortunate thing is that (unless you manage to utterly break it) no matter what's being tested, thing 'worx' and measurements ~random
signpost[asciilifeform]: would like to end up with something chopped down to nothing but eating and farting blox
asciilifeform wonders how many times the thing could've been rewritten in the cycles folx have spent in attempts to 'civilize'
awt[asciilifeform]: Will be nice to be able to paste exmandable code snippets in here
dulapbot: (asciilifeform) 2022-03-20 asciilifeform: the ~complete lack of a detectable interest in e.g. trb among 'mega-hodlers' already speaks volumes re their intelligence.
asciilifeform found, in recent yrs, that attempts at anyffin beyond ultra-simple convenience pheature in re trb rapidly goes to quagmire of rapidly diminishing returns
dulapbot: (asciilifeform) 2022-02-17 asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2022-02-17#1079578 << prolly won't surprise cgra that asciilifeform concluded that doing ~anything~ in parallel in trb is ~impossible w/out removing locks and potentially opening gates of hell. asciilifeform's draft 'usecement' is 100% serial, asks for 1 block at a time (and wastefully, 'getdata' to each connected peer, thinking atm how to reduce the bw guzzling)
asciilifeform: while on this subj, tbf piggy ftr to this day precisely as it was in '21 . 'no +/- planned.'
dulapbot: (asciilifeform) 2021-11-11 asciilifeform: for reference, piggy, seen through a heathen blockexploder .
dulapbot: (asciilifeform) 2021-12-18 asciilifeform: ... problem is, is very difficult to increase the amt of useful work with coin (aside from, hypothetically, jailbreaking ~people~, ideally 2-3 people. but piggy -- or 'sandwich-adjusted' exchrate -- would have to be ~10x bigger to make this plausible)
signpost[asciilifeform] not planning mega-refactors. walletless patch was a circa hm, 2019 item? possibly earlier.
signpost[asciilifeform]: in dusting that off, poked around and wondered wtf the semantics were for inventory management of blocks.
signpost[asciilifeform] for his own purposes only needs the sendrawtransaction hole he bored in the side.
asciilifeform: signpost: how about in the other direction ? ( how does the 'wallet' apparatus find out about unspent outs ? )
signpost[asciilifeform]: there's that transaction index the wallet uses that I need to expose via RPCs
signpost[asciilifeform]: haven't dug into it yet.
asciilifeform: possibly 'wallet' oughta swallow blocks (from local noad) . this way 0 leakage from 'wallet' to outside in re what addrs are of interest
asciilifeform: ( iirc there was a thread about this, but can't seem to find in o(1) )
asciilifeform: ( wallet not needs to ~store~ the blocks whole, as it never needs to send'em anywhere, but oughta sanity-check min-work etc., store hashes, confirm that it's being fed valid blox, & watch for the req'd unspents )
asciilifeform pictures 'wallet' as physical apparatus that oughta work w/out 'special' soft support on pc end :
asciilifeform: ... 2 photodioded serial ports ; 1 can eat sequence of new blox (at convenience of operator, whenever he has reason to think he's been paid, or earlier has spent sumthing ) ; other end output-only, can emit a tx. and on front panel , knobs for selecting unspent outputs / amt / destination .
asciilifeform: the tx oughta be droppable in trb or even the various public toilets where paste.
asciilifeform: networked noad has 0 'need to know' what addrs you find 'interesting' unless yer spending.
asciilifeform: on this hypothetical iron, the 'knobs' oughta include a means of entering a decryption pw. the keys per se, to live on removable flashes, 'unclassified when unkeyed'(tm)(r) ergo as many backups as one feels like
asciilifeform: ... to the flash stick, write ciphered keys not as file but as luby into N unused-by-fs blocks.
asciilifeform: then use stick for civilian purposes, taking care not to fill it above ~30% or whatnot.
signpost[asciilifeform]: yup, nicely avoids chug associated with particular blocks leaking a bit of timing
asciilifeform lulz over popular 'iron wallets' which are readily identifiable to naked eye as 'crown jewels' and contain plaintext keys (in 'unbreakable'(tm)(r) ic)
signpost[asciilifeform]: lots of items I'll be happy to work on if the definitive sovereign bids.
asciilifeform: signpost: not simply re timing, but no networked machine has any biz knowing what 'your addrs' are, or that a tx it is given 'is yours'
signpost[asciilifeform]: !!ticker btc usd
signpost[asciilifeform]: meanwhile, got as far as check block generation with "occ"
asciilifeform: occ ?
signpost[asciilifeform]: "online codes c"
signpost[asciilifeform]: http://paste.deedbot.org/?id=TbrL << approx 345MB/sec single-threaded, looks like
signpost[asciilifeform]: way fucking better than what I was getting out of py, or my probably really crappy lisp.
signpost[asciilifeform] not the wizard that can coax mega-optimizations out of sbcl
signpost[asciilifeform]: decoder will be more of a bitch.
asciilifeform: decoder is necessarily 'over9000' moar expensive than encoder, for these
signpost[asciilifeform]: yup, of course. shared state of which blocks already solved, etc
asciilifeform spent far too many cycles chasing 'perpetuum mobile' of trying to devise a lubyesque decoder that could somehow identify bogus chunks
asciilifeform: i.e. where 'no credit extended' to counterparty
asciilifeform not found even probabilistic algo for any such thing, to date
signpost[asciilifeform]: interesting thought, would be nice for large swarm downloads where not all participants are known in advance.
signpost[asciilifeform]: you at least get the progressive attenuation of wot-tronic trust by being inside pestnet.
signpost[asciilifeform]: direct peers, I'm most certainly going to try to decode their blocks, and if I really want the file, might at a specific point in time accept streams from L>1.
asciilifeform: would be nice for all kindsa things.
asciilifeform: but asciilifeform not knows whether possible even in principle.
asciilifeform not expects anyone to bake sane 'iron wallet' -- the 'how' was imho obv. for 10+yrs nao, but who would pay
bitbot[asciilifeform]: Logged on 2023-05-24 14:39:28 asciilifeform: on this hypothetical iron, the 'knobs' oughta include a means of entering a decryption pw. the keys per se, to live on removable flashes, 'unclassified when unkeyed'(tm)(r) ergo as many backups as one feels like
asciilifeform: ... fwiw 'sane wallet' could even store all the blox ( 1tb of ssd is ~50bux last asciilifeform saw ) and so be able to swallow arbitrary privkeys and find the unspents