320300+ entries in 2.322s
mod6: so can we rewind the tape just
a bit...
mircea_popescu: sipa jgarzik: have we seen
a block which affected 5000 transaction index entries?
benkay: you mentioned something about the dev team starting work on
a spec?
benkay: it's
a transient state. frustrating to be sure.
Anduck: and pay
a huge fee of.. 0.1 btc!!
mod6: ok i gotta walk away for
a bit and think about this some more.
jurov: mod6 i'm not interested in arguing. if i had time, i'll do
a fork with some very useful function and wait for them to come
kakobrekla: oh i read that "in summary the coinbase is defenetly useful" and got
a heart attack almost
mircea_popescu: mod6 actually once you're done with the iphone app,
a very good use of someone's time is to go through the codebase and summarize it
MJR_: I think
a central clearing corp, that only needs to transfer the diff between exchanges will be the best way
mod6: jurov: we can learn
a thing or two, perhaps, about how DNS solves the issue. UDP is meant to be fast, and arbitrarily large since forward/reverse zones can contain an large amount of entries.
MJR_: I wrote
a program to monitor how often we need to fail over to TCp for our data feeds
mircea_popescu: in
a sense mpex is constructed exactly opposite of how bitcoin is constructed.
mod6: and really... the UDP (TCP when needed) is
a model that the guys should be looking at to solve this blocksize issue
mod6: not the SAME code, but similar to avoid attacks upon
a weakness found in one
MJR_: I really think
a gpg based order system would be nice
mod6: if they were modeled after mpex, that'd be
a start
MJR_: anyway I think there is
a ton of room for competition to mt gox
MJR_: they listed the loss in dollars, but you can't send cash over
a wire
pgp: you can lead
a horse to water
pigeons: mjr_ where have you been? bitininstant lost usd
a few days ago
mircea_popescu: what's sad is that they take ~5k
a month in hardware costs to process 8k trades in 30 minutes
mircea_popescu: problem is every exchange wants to act as if in
a vacuum, because
mircea_popescu: or independently, but pgp is
a good solution to keep it safe.
pgp: this would also require something akin to DTCC (depositiory trust clearing corp) where every exchange is
a member and keeps money on deposit...
MJR_: well, again I think that could happen on
a larger scale...
mod6: and fixing this block size thing is kinda crazy. its like you know when you get
a UDP request to your DNS server and the reply from the Zone is > than the max size of UDP, it rolls over to tcp
mircea_popescu: Mar 11 09:19:27 <mircea_popescu> this process is called fixing, it's widely used to obtain
a price.
mircea_popescu: Mar 11 09:19:08 <mircea_popescu> now
a b and c review their books
mircea_popescu: Mar 11 09:18:58 <mircea_popescu> x has to call
a lower price now
mircea_popescu: Mar 11 09:18:45 <mircea_popescu> then they say.
a buys 50k usd. b sells 25k usd. c sells 70k usd.
mircea_popescu: Mar 11 09:18:27 <mircea_popescu>
a, b and c go through their books, see what customers want, see what they want on their own account
mircea_popescu: Mar 11 09:18:00 <mircea_popescu> ok, it's like this. suppose the three main dealers are
A, B and C.
mod6: root servers don't keep the same code either, not exactly anyway, this way it prevents
a simutaious attack against all
mod6: BTC needs
a CVE or something or some sort of post mortum.
mod6: yeah BDB has been
a thorn in dev's side for
a decade. but sounds like this time they forgot about
a landmine they already stepped on once before.
pigeons: you can configure it to use more locks, so far that seems to work just uses
a lot more ram but works
mircea_popescu: pigeons i fear it'll take me
a little to digest this one
mircea_popescu: this is possibly the worst case of "hi i am calling to report
a bug" bungled tech support ever.
pigeons: "If anyone has any idea how I can buy
a large amount of bitcoins please message me. The largest amount I can buy from bitinstant is $500 worth ive looked and looked and its near impossible to make
a large bitcoin purchase anywhere."
pigeons: "if I was
a scammer I could be living it up easy some place warm instead of braving the same upstate new york winters I have been for 30+ years"
pigeons: today I downloaded .7 and have been patiently waiting for my blockchain to download, im not sure if I will continue refunds or just buy
a big crack rock and smoke it in
a cave some where but I am totally offended by this scammer tag
mircea_popescu: until someone figures out how to trigger the berkley bug with
a smaller block
iz: ThickAsThieves: *picture of xibit holding
a picture of xibit*
Luke-Jr: imsaguy: draw
a big circle and write 1.0 in the middle?
imsaguy: if they cut people off cold turkey, they could cut
a lot of hackjobs added to make old software work with the new stuff
smickles: where's the statics on what version people are running, i think there are
a bunch still on .3
smickles: mircea_popescu: would
a better measure be sending
a bunk order?
ThickAsThieves: so i know they me seem ignorant, but just thinking out loud, is there not
a way to make the bitcoin software part of the blockchain itself?
mircea_popescu: the lag to talk to mpex, through
a remote and
a local proxy
benkay: centralize us on
a single buggy broken blatform
mircea_popescu: knowing in advance there will have to be
a hard fork is
a weakness of the system.
Namworld: No, I mean malicious stuff from knowing ahead of time about
a fork.
ThickAsThieves: so does that not give malicious people
a specific time in the future to plan for?
Namworld: and devs can consult with mining pool when changing limits and send
a warning so everyone upgrades directly to the client, should they want to make
a change to allow scaling.
mircea_popescu: this seems to be the contemporary brain disease, people reading stuff they don't understand here and there, mashing it into
a sort of compound
imsaguy: sounds like
a methlab waiting to happen
Namworld: If it was
a written text, I might make less omissions. When chatting tho, I type fast and usually omit things.
Namworld: I think you are very literal on what sentences mean and interpret as little as possible. You avoid making too many inferences during
a conversation so even small omissions can result in misunderstandings.
jborkl_: The problem glaring from this, now
a obvious major problem is here and can be exploited again.
Namworld: Actually I find you go about things in
a different way than anyone else...
mircea_popescu: which is
a great way to become
a laughingstock, sure, but also
a great way to find obscure bugs.
mircea_popescu: i enjoy arguing with him mostly because he seems to go about things in
a totally different way than anyone else
jborkl_: They have identified it as .BDB has
a limit and can be broken. .8 does not have
a limit and seems fine. Lets move back to the broken
Namworld: I hate it when I forget something in
a sentence and I'm getting argued against for 10 minutes when I have the same point of view as the one arguing against me.
smickles: so it's
a silly argument :P differences in definitions
MJR_: Namworld: I think what they mean is that part of testing
a new release is testing it's interactions with other versions for backward compatiblity
jborkl_: I just solo mined on .72 again last night until it was all figured out- I knew I kept .72 around and running for
a reason
mircea_popescu: nobody is saying "shouldhave been tested more by itself in
a vacuum"
jborkl_: really, only test
a part of the code against itself?
smickles: he only tested 500k blocks, and it happened on
a 900k block
smickles: Namworld: gavin said he would've caught it if he had tested
a larger block on the testnet
jborkl_: I ran .7 and .8 versions, If I had run testnet and made
a 1mb block I would have most likely broken .7
mircea_popescu: any statement that claims "no amout of testing" is
a lie by its nature
mircea_popescu: in
a sense this backhandedly proves how strong the idea is
smickles: Troic_: beauty of the system, they have
a financial incentive to cooperate
jborkl_: That they encouraged people to change the block size and not knowing it was going to break BDB was
a big fuck up
Namworld: It wasn't that colossal... just
a little forking
jborkl_: It is holding up well considering what
a collosal fuck up this was
midnightmagic: More likely it was
a side-business for coin laundry, but none of the miners could ever (far as I know) figure out where the coins were coming *from*.
mircea_popescu: is the main strategic purpose of having
a devteam in the first place.
mircea_popescu: just leave it stand as it is. the theoretical response to
a 51% attack is that the attacker would make more by mining.