< chatter29>
to accept Islam say that i bear witness that there is no deity worthy of worship except Allah and Muhammad peace be upon him is his slave and messenger
< bitcoin-git>
[bitcoin] AllanDoensen opened pull request #10156: Fix for issues with startup and mutiple monitors on windows. (master...masterMultiMonFix) https://github.com/bitcoin/bitcoin/pull/10156
< bitcoin-git>
bitcoin/master cd7f394 John Newbery: initialize flag variable to 0 (and continue if GetLogCategory() fails)
< bitcoin-git>
bitcoin/master c7e73ea Wladimir J. van der Laan: Merge #10151: [logging] initialize flag variable to 0 (and continue if GetLogCategory() fails)...
< bitcoin-git>
[bitcoin] laanwj closed pull request #10151: [logging] initialize flag variable to 0 (and continue if GetLogCategory() fails) (master...logging_initialization) https://github.com/bitcoin/bitcoin/pull/10151
< achow101>
weren't there some memory usage improvements in 0.14.1?
< BlueMatt>
achow101: for default configs, yes, and slightly reduced in general from 0.14.0, yes?
< achow101>
Ok. Well I think they should be documented in the release notes
< cfields>
I'm working on building/signing rc1. I'm out of town and using my laptop, so it's a bit painful and slower than usual.
< bincap>
when writting functional tests in python, we can use .generate to add blocks to test blockchain; and can I add a block with manually set flags, like segwit signalling on/off?
< bincap>
also, how to set given date of the block generation? I want to test bip148 that rejects blocks that do not signal for segwit activation
< bitcoin-git>
[bitcoin] jnewbery opened pull request #10159: [tests] color test results and sort alphabetically (master...sort_test_results) https://github.com/bitcoin/bitcoin/pull/10159
< jeremyrubin>
gmaxwell: I think your asicboost description is backwards?
< jeremyrubin>
or am I misunderstanding
< jeremyrubin>
distinct first 64-byte chunks but identical 16-byte
< jeremyrubin>
Oh maybe I'm misunderstanding how asicboost works?
< gmaxwell>
I think you were misunderstanding, or my explination is unclear.
< gmaxwell>
for asic boost you need multiple different first-compression runs for each second compression run. This allows you to reuse the computation for the message expansion in the second compression run.
< gmaxwell>
I think part of the reason people haven't taken it seriously as a threat is because it's tricky to understand.
< gmaxwell>
while talking about it privately for the last month I ran into many experts that kept lapsing into thinking that the collision required 2^32 work, and other misunderstandings.
< jeremyrubin>
gmaxwell: I think it gets a bit scarier if you don't include any transactions?
< jeremyrubin>
gmaxwell: Maybe this is related to why ------- is mining empty blocks?
< jeremyrubin>
gmaxwell: or if you only include stale transactions you control and guarantee won't be spent in a prior block
< jeremyrubin>
gmaxwell: this would suggest your solution of including header hash is worth it
< gmaxwell>
My goal here is only to block the form which is incompatible with protocol improvements like segwit that add additional per transaction data. I leave blocking asicboost more generally an open question. A miner could mine empty blocks too, but they'll at least be disadvantaged by the loss of fee income-- so, personally, I feel that is largely self correcting.
< jeremyrubin>
I see; I think that's reasonable.
< jeremyrubin>
You don't nescessarily lose that much fee though this way, I'd point out. You can just mine things closer to the bottom which are unlikely to get mined.
< jeremyrubin>
You can also mask it by mining a full block with only your transactions in it.
< Taek>
even masked, you are still disadvantaged
< jeremyrubin>
Taek: yeah, but you get a huge benefit, because your set of valid RHS commitments goes up over time and only decreases (by 1) when you mine
< gmaxwell>
jeremyrubin: if you can't do the left/right split trick you have to do a log(txn) hashes for each try.
< gmaxwell>
In any case: the goal, again, isn't to stop asicboost but to stop forms that are incompatible with protocol extensions. I believe anything that works after this bip works just as well with segwit and friends.
< gmaxwell>
I expect other people to make proposals to block asicboost more completely (in fact, I kind of cut off private discussion along these lines with this proposal-- because I think it's critical to seperate the concerns)