< wumpus>
MarcoFalke: #20170 seems to be stuck on your signature; is there anywhere we can find a non-expired version of your PGP key? it fetches the trusted keys from the ubuntu server as the CI does but somehow that doesn't work (not locally either)
< sipa_>
input/output errors means hardware failure
< S3RK>
you're right. I tried to remount disk and then Deserialize or I/O error - CAutoFile::read: fread failed: unspecified iostream_category error at FlatFilePos(nFile=2479, nPos=69638818)
< vasild>
MarcoFalke: wrt fuzzing, right, that is my understanding too. And if a test is changed (very little) to just do ConsumeBool() before any other consumes, that would "invalidate" the fuzz seeds, I guess.
< MarcoFalke>
vasild: integral types smaller than 2^64 are consumed from the end. Byte vectors from the beginning.
< MarcoFalke>
So it could happen that adding a ConsumeBool doesn't invalidate some seeds
< vasild>
I see
< MarcoFalke>
Currently it looks like we are hovering around 64%-65% total coverage. Decreasing when the code changes, increasing when the fuzz engine caught up
< vasild>
I wonder if I should try to create a seed _manually_, because it is known that a reply on the socket like "foobar RESULT=OK baz" would trigger more coverage
< vasild>
but waiting for the fuzzed to discover this may take a looong time
< vasild>
s/fuzzed/fuzzer/
< darosior>
vasild: manually generating a corpus is generally a pretty good idea (can't think of a scenario where it would not?), so that the fuzzing engine can start from this to discover new code paths
< vasild>
alright, will try that
< MarcoFalke>
You can also try to embedd a dictionary in the source code (The message type fuzzers do that with the message type)
< MarcoFalke>
If the replies are "OK" or "NOT_OK", you could generate them as: reply = ConsumeBool()? Pick({"OK", "NOT_OK"}) : ConsumeString()
< MarcoFalke>
Ideally the fuzz engine would figure out all of this by itself, but we are not there yet
< bitcoin-git>
[bitcoin] MarcoFalke opened pull request #21388: scripted-diff: [doc] Rename fuzz seed_dir to corpus_dir (master...2103-docInputDir) https://github.com/bitcoin/bitcoin/pull/21388
< Norrin>
dimitaracev35 i think here would be better for that
< dimitaracev35>
Ah, already asked that question here but chat seems to be quiet atm
< dimitaracev35>
thanks anyways!
< Norrin>
dimitaracev35 that's weird. it's not in my scrollback
< dimitaracev35>
it was under a different username since I'm quite new to irc and had to make an account with weechat
< bitcoin-git>
bitcoin/master 51e2cd3 Andrew Chow: Have CalculateMaximumSignedTxSize also compute tx weight
< bitcoin-git>
bitcoin/master 3e69939 Andrew Chow: Fail if maximum weight is too large
< bitcoin-git>
bitcoin/master 48a0319 Andrew Chow: Add a test that selects too large if BnB is used
< bitcoin-git>
[bitcoin] meshcollider merged pull request #20536: wallet: Error with "Transaction too large" if the funded tx will end up being too large after signing (master...fundtx-error-large-tx) https://github.com/bitcoin/bitcoin/pull/20536
< jonatack__>
exit
< luke-jr>
O.o
< jonatack>
(oops)
< bitcoin-git>
[bitcoin] achow101 opened pull request #21392: Implement BIP 8 based Speedy Trial activation (master...bip8-speedy-trial) https://github.com/bitcoin/bitcoin/pull/21392