--- Day changed Mon Oct 22 2018 00:02 < andytoshi> i know, it's horrible 00:02 < andytoshi> what else are we supposed to do with a sunday evening 00:03 < BlueMatt> yeesh wtf 00:03 < BlueMatt> fucking microsoft 00:04 < BlueMatt> ariard: do you have the input that crashed the fuzzer? 00:06 < ariard> heey how do we extract it ? it's in hfuzz_workspace/full_stack_target/input/ ? 00:07 < BlueMatt> were you running the travis-fuzz script? 00:08 < BlueMatt> it should have cat'd the file for you 00:08 < ariard> at root of repo 00:08 < BlueMatt> otherwise look at hfuzz_workspace/full_stack_target/ 00:08 < BlueMatt> files named SIG* 00:14 < ariard> well don't find anything with SIG* pattern 00:15 < ariard> I still have the inpust in hffuzz_workspace/full_stack_target/ 00:18 < BlueMatt> there's no files in hfuzz_workspace/full_stack_target/? 00:18 < BlueMatt> what was the crash message? 00:19 < ariard> crash at 764 in full_stack_target 00:19 < ariard> with None instead of Some 00:20 < ariard> failure of test_no_existing_test_breakage 00:22 < BlueMatt> ohohoh, not a fuzz failure, a fuzz-crate-test-failure 00:22 < BlueMatt> wait so cd fuzz && cargo test failed once but not multiple times? 00:22 < BlueMatt> the original branch def failed there 00:22 < BlueMatt> but I fixed it 00:22 < BlueMatt> it passed travis, but you saw that failure on the latest copy? 00:22 < BlueMatt> I mean I fixed it like 30 minutes after creating the pr 00:23 < ariard> well i saw it with the current state of your branch 00:24 < ariard> usually when I review PR, I launch the travis tests with cd fuzz/ ; cargo test && travis-fuzz.sh 00:24 < BlueMatt> hmm, weird, that *definitely* should be deterministic 00:25 < BlueMatt> if you run fuzz's cargo test in a loop do you see it? 00:25 < ariard> I've launched the test multiples times after it but seen only once 00:26 < BlueMatt> 'tf 00:26 < ariard> with fuzzing, input aren't randomly generated at every round test ? 00:29 < BlueMatt> not the fuzz/ cargo test 00:29 < BlueMatt> honggfuzz sure, but the failure you saw there was cargo test 00:33 < ariard> well I'm running this in a loop, don't see it again crashing right now 00:36 < BlueMatt> ugh :( 00:41 < ariard> BlueMatt : got it again 00:42 < ariard> just running while [ $? = 0 ] ; do cargo test ; done for enough time 00:43 < ariard> so yes sorry it's isn't a honggfuzz crash 00:44 < ariard> thread 'tests::test_no_existing_test_breakage' panicked at 'assertion failed: `(left == right)` 00:44 < ariard> left: `None`, 00:44 < ariard> right: `Some(1)`', fuzz_targets/full_stack_target.rs:762:3 01:25 < BlueMatt> ariard: ahah, yea, its probably a rng reset race across the two tests... 01:26 < BlueMatt> should exist already on master, I think 01:26 < BlueMatt> i guess should disable the other test in that file 01:26 < BlueMatt> that should fix it 01:26 < BlueMatt> given the other one is a dummy test 01:30 < BlueMatt> if you comment out the other test it should be reliable 01:45 < ariard> Ah okay, it seems that was the good reason, running for 10min and no crash 01:52 < ariard> BlueMatt: finish to review #217, seems fine to me, no issues to comment, maybe few nits but nothing jamming 01:52 < ariard> hope github will be resurrected tomorrow 09:50 < stevenroose> Sorry for the 5 exact same comments, GitHub was down and I kept trying :D 10:44 < stevenroose> andytoshi: the bip32 impl in rust-bitcoin doesn't have a Path type representing a path like `m/44'/0'/0` etc right? Just `Vec` I suppose? 11:53 < TamasBlummer> see rust-wallet project, keyfactory.rs bip32_tests 11:53 < TamasBlummer> if you'd like it simpler 12:31 < stevenroose> TamasBlummer: aha, thanks! Never looked at rust-wallet actually 13:18 < andytoshi> correct, rust-bitcoin just uses Vec 14:15 < BlueMatt> hey TamasBlummer! 14:16 < BlueMatt> how's rust wallet stuff going?! 14:16 < TamasBlummer> hey @BlueMatt 14:16 < TamasBlummer> have been working on the blockchain store for the last monts 14:17 < TamasBlummer> the wallet needs that so it can scan adresses 14:18 < TamasBlummer> the blockchain store is quite advanced. I think it will beat the combination of flat file and leveldb such as e.g. BlockSci 14:18 < BlueMatt> oh? I thought it was an spv wallet? doesn't it just need headers + list of txn that you care about? 14:18 < TamasBlummer> the spv is also stalling on the same 14:19 < TamasBlummer> I want to store block filters for spv, that also needs a good db 14:20 < TamasBlummer> I admit that I started quite a few things and did not yet finish one. but am close, really for the block store 14:22 < BlueMatt> ahah, so you want to store the full bip 157 filters per-block? 14:22 < TamasBlummer> yes and the blocks also if onw wants 14:24 < BlueMatt> ah, but only optionally in the wallet, no? 14:25 < BlueMatt> I mean thats a lot of data just to have rescan ability which you may only sometimes want? 14:25 < TamasBlummer> these are all building blocks, nothing is a must 14:25 < BlueMatt> coolcool 14:25 < BlueMatt> thats awesome 14:25 < BlueMatt> glad things are moving along =D 14:27 < TamasBlummer> sorry for being offline quite often. I have been hospitalized a few times in the meanwhile and have a very unpredictable working schedule and mood 14:31 < BlueMatt> this is a volunteer set of projects! you cant apologize for that! 14:31 < BlueMatt> sorry to hear about the health issues, hospitalization is never any fun :( 14:34 < TamasBlummer> Thanks , no worries. I am in best hopes both for my health and for finishing what I started. Let me know if I create some dependency. 14:36 < BlueMatt> nah, I think we've done a pretty good job of everything running independently from the start, so dependencies aren't too much a concern 14:37 < BlueMatt> if anything you'll end up blocked on rust-lightning's api stabilizing more than the other way around :p 15:01 < stevenroose> dongcarl: just noticed something about Rust when using the PSBT API :D when you have a type (K, (V1, V2)) and you want the v2 object and you do `something.1.1` it sees `1.1` as a float and fails :| 15:02 < TamasBlummer> (something.1).1 works 15:15 < stevenroose> TamasBlummer: yeah it does. Just funny quirk. 15:16 < stevenroose> andytoshi: any particular reason why there is no `From for &[u8]` and `From for Vec`? I've missed them several times. I think it's weird to have the consensus_encode stuff on it as the only serialization because hashes are quite more general 15:19 < TamasBlummer> see Sha256dHash::from_data and Sha256dHash::data 15:24 < andytoshi> stevenroose: the From story with Sha256dHash is a mess right now 15:24 < stevenroose> andytoshi: oh there is to_bytes. Was looking on wpsoftware.net.. 15:24 < andytoshi> we have From<&[u8]> which i think panics for a non-32-byte input, which is insane .. and then no From for anything .. 15:24 < stevenroose> to_bytes is perfect. Can do to_bytes()[..] 15:25 < andytoshi> o.O i had no idea the wpsoftware rustdocs were still up 15:25 < andytoshi> i guess they're just static html + cron, there's not really any way they could've gone down.. 15:25 < stevenroose> Yeah the two Froms would make sense imo. If you agree, I can make a PR. 15:25 < andytoshi> no, that should all go in bitcoin_hashes 15:25 < andytoshi> i don't think we should do anything with the existing hash types in rust-bitcoin 15:26 < stevenroose> Oh, they are being moved to bitcoin_hashes? As a rust-bitcoin dependency? 15:26 < andytoshi> yes 15:26 < stevenroose> What's the reasoning for that? 15:26 < andytoshi> they're useful for more than just bitcoin 15:26 < andytoshi> and right now we have a dep on rust-crypto which is breaking a bunch of people 15:26 < andytoshi> so we have to replace that 15:28 < andytoshi> ugh, github is still all fucked up 15:28 < andytoshi> why don't they take the freeking site down instead of having random 404s and timeouts 15:30 < TamasBlummer> wont get better with microsoft 15:31 < andytoshi> evidently not 15:32 < Blackwolfsa> @BlueMatt I made have one or two questions in PR #202 regarding the return type of the trait function, and the storing of raw messages, I left comments as to way I did it that way, but I am happy to change them around. 15:38 < BlueMatt> TamasBlummer: I kinda figure they're moving it to azure and thus its going to shit lol 15:39 < BlueMatt> Blackwolfsa: I'll take a look in a bit, trying to rebase and get 157 merged so that I can rebase+merge the big ones of mine without completely breaking it 15:39 < BlueMatt> I dont think 202 is as bad rebase-wise for things like 217 21:54 < stevenroose> github going down is one thing. I've been trying to book flights with vueling since this afternoon (23:00 here now) and since I started, tix have gone up from 280 to 310 EUR. Giving me timeout errors