153500+ entries in 1.066s
decimation: his code sucks, I don't deny, but it's the closest thing to
a spec that we have
midnightmagic: .. which appears to have been
a gavin/hearn originated bug.
decimation: again, nobody is forcing anybody use
a version of openssl they don't want to use.
midnightmagic: or, do what sipa did and write
a secp256k1 lib because the openssl people don't give
a shit they're wrecking dependencies.
decimation: pretending like others are there to solve your problems seems like
a poor approach
midnightmagic: well, then we have the forking risk I mentioned above: what happens when the openssl people make
a fix or the internet finds
a bug in the component that we depend on? If we sit on 0.9.8 or whatever the version was before those idiots got their hands on it and started adding malicious exploits, what happens?
☟︎ ben_vulpes: <decimation> why wouldn't we want
a static bitcoind that is correct for all time? << we do. midnightmagic does not.
decimation: why wouldn't we want
a static bitcoind that is correct for all time?
midnightmagic: I'm familiar with the DER-encoding change they made, and I'm aware of, if not familiar with, every major bug in openssl since 2001 or so. Could I draw
a line between releases that had bugs and releases that fixed them? No. Not even close.
☟︎ decimation: midnightmagic: who is holding
a gun to your head, requiring you to update openssl?
midnightmagic: Thus, backing up further, there is the legitimacy of BIP66 at all. If it is not legitimate, we have
a consensus code failure every time openssl decides they want to change behaviour.
midnightmagic: ben_vulpes: On that at least, we agree. I agree with that: the current hashrate is as illegitimate as
a vote of private keys would be in determining
a softfork. What else is there?
midnightmagic: Regardless,
a 1000-block window is not unreasonable if one accepts that mining hashpower is the vote that counts.
midnightmagic: ben_vulpes: You would call
a bite any response. That is the fundamental nature of #trollfail.
ben_vulpes: since when has midnightmagic's employer given
a shit about profitability?
ben_vulpes: efficiency is not
a prerequisite for fab runs.
decimation: also, it's probably not going to be true in
a year or two as 14 nm asics fan out and become barely economic
ben_vulpes: midnightmagic: this is
a basic feature of integrals and curves.
midnightmagic: i'm just saying that it sounds like
a little bit in comparison to both time, and integerial block height, but actual work-wise it
a significant chunk. again, very unfortunately.
decimation: to spite, you could even pick
a value between 4 and MAX_INT randomly
midnightmagic: decimation: 1000 blocks is
a couple percent of all work done ever on the entire blockchain since inception, and the current hashrate could rewrite the entire history up to something like august 2014 in somewhere close to the span of time that non-vote took place over, times
a very small number. :( unfortunately.
decimation: one week out of years of doing things
a certain way
decimation: I'm fairly certain that if 0.5.3.1 were used to mine
a block with nVerion=1 it would be rejected
ben_vulpes: "reasons" being that i wrote
a blog post saying "i don't think much happened after this"
ben_vulpes is amused, but not honored. if ye'd only had actual management once upon
a time, this'd not be
a notable achievement.
mod6: s/must be considered/is
a/ ?
decimation: I would say that on its face, bip66 isn't such
a terrible idea
☟︎ ben_vulpes: perhaps
a hash of the codebase instead of the semver crap?
ben_vulpes: mircea_popescu, mod6, asciilifeform: i don't see
a reason to make this much more than 25 words. do you think much detail or polemic is necessary here?
decimation: 'you may not be interested in
a fork, but...'
ben_vulpes: this is almost as bad as defining
a word with the word to be defined
ben_vulpes: i thought you were running
a conformal impl somewhere
mircea_popescu: but the "we'll make wallet
a plaintext file" and the "we'll target windows" and the "you know what's better than boost ? qt!" were pretty epic turn points too
ben_vulpes: i want
a man in the ground, i don't make that dependent on his wife squirting or w/e
ben_vulpes: hey kakobrekla how much can i cram into
a rating field?
mircea_popescu: why ? he's
a god fearing rotinculo from wisconsin or whatever.
decimation: because they already did
a 'softfork' on the nVersion
mod6: <+ben_vulpes> Bitcoin 0.5.3 is the canonical reference implementation. If
a fork occurs and one side validates on the 0.5.3 codebase while the other does not, the chain that validates under 0.5.3 is the only valid chain. << I have nothing further to add to this at this time.
mircea_popescu: i have nfi how the power rangers imagine they'll manage
a hardfork when they can't as much as get
a "relevant support" softwork that's fairly uncontroversial.
ben_vulpes: i have
a porch of meatwot and babes that actually need attending to.
mircea_popescu: anyway, todd is right. the only way this affects us is that we don't really give much of
a shit.
mircea_popescu: Bitcoin Core (after 0.10.0) rejects these invalid blocks, but
a lot of other stuff doesn't. SPV Bitcoinj wallets do no validation what-so-ever, blindly following the longest chain. blockchain.info doesn't appear to do validation as well; who knows what else? <<
ben_vulpes: Bitcoin 0.5.3 is the canonical reference implementation. If
a fork occurs and one side validates on the 0.5.3 codebase while the other does not, the chain that validates under 0.5.3 is the only valid chain.
mircea_popescu: anyway, foundation ppl : plox to make
a statement on this matter
ben_vulpes: my position being "if your notion of
a valid block has to patch 0.5.*, get fucked."
phf: somebody was saying bitcoind doesn't build on
a 32 bit system?
decimation: yeah I think we were forked
a long time ago actually
mircea_popescu: <ben_vulpes> of the active users. << well, like
a dozen or so people, spent
a month in game last month. so like, 2 hrs / day or somesuch
mod6: ah, i can give it
a go with something as old as like 4.5.4 for sure.
mod6: any idea how old of
a version is required?
mod6: <+asciilifeform> mod6, ben_vulpes ^^^ who wants to try << i might be able to give it
a go later this weekend ... maybe.
decimation: asciilifeform: I've tried printing such things before, even on
a big plotter the results are disappointing
decimation: asciilifeform: maybe the naked blocks are
a counterpoint :)
decimation: the big blocks are running nearly
a mb or so
decimation: it would sure make it simpler to devise
a big asic machine to find hashes if it doesn't need to bother with transactions
decimation: I guess as long as they find
a block that matches difficulty it will be valid
decimation: I suppose if
a peer is available, you could do that. but then you are only trusting one node's say
decimation: it would be useful to know when
a new block is found, so you can ask your peers for
a copy
decimation: for the headers thing, I'm imagining
a scenario where someone is waiting for
a transaction to clear
decimation: aye, that would be useful for
a 'starter pak' node
decimation: it occurs to me that it would be useful to broadcast
a digest of
a block in addition to the actual block data
decimation: asciilifeform: did we do
a thread on 'headers only' block data?
punkman: ok, stator compiled successfully I think, anyone wanna give me
a node IP?
trinque: all the events which transpired which are marked by
a particular transaction
trinque gets
a funny feeling like he's watching
a fast-forward replay of history itself whenever syncing
a new blockchain
phf: the general pattern is
a bunch of read-* functions, that are all ultimately reduced to read-byte or read-sequence. the an (unsigned-byte 8) stream and return whatever datrastructure