--- Day changed Tue Nov 13 2018
00:36 < dongcarl> andytoshi: Where are you tracking the reproducibility of rustc for debian? I don't see many successes here: https://tests.reproducible-builds.org/debian/rb-pkg/unstable/amd64/rustc.html
01:45 <+andytoshi> dongcarl: woot
01:46 <+andytoshi> and i haven't been tracking reproduceability, i've just been taking debian as "the best that exists" as far as reproduceability
01:46 <+andytoshi> and less specifically, as "much more trustworthy than rustup.rs or wherever the rustc devs are uploading to"
01:47 <+andytoshi> though that's a pretty low bar :) i trust the centos packagers more than rustup but i still don't think they're looking much at what they're packaging
01:56 < dongcarl> Ah, gotcha
01:57 <+andytoshi> Aside from being pretty gung-ho about determinism (i'm sad they haven't succeeded) i like that debian has a working WoT and a culture of responsible package maintenance
01:58 <+andytoshi> in the morning we can talk about upgrading stuff to rustc 1.24. i don't want to gratuitously increase versions just because debian has moved .. in general we want to try to support the oldest rustc we can
01:58 <+andytoshi> but like, 1.15 gives us custom #derive which we really want for all the serde stuff .. plus it gives us serde_json support which we want for rpc
01:59 <+andytoshi> i'm not sure when associated constants arrived
01:59 < dongcarl> Yeah... we can discuss it then... associated constants are what I'm most excited for
01:59 < dongcarl> It's before 1.24
01:59 <+andytoshi> but those woud make the bitcoin_hashes story much better
01:59 <+andytoshi> ah, it's 1.20
02:00 <+andytoshi> so yeah .. lemme think a bit more about it .. but i think upgrading to 1.20 is probably worth the breakage
02:05 < dongcarl> :+1:
02:07 <+andytoshi> i'm a bit apprehensive about 1.31 ... i think there's gonna be a lot of people upgrading to rust 2018 cuz they're excited about it, and will make a bunch of breaking changes for no good reason
02:07 <+andytoshi> though i guess most all of our deps are by groups who've been around a long time and who care about breakage
06:20 < BlueMatt> stevenroose, ariard: hey, just got back from hiking for a few days
06:21 < BlueMatt> stevenroose: rust-lightning has as much p2p as its ever gonna have (that is to say, it has a full p2p stack, but does not do the actual socket calls, you have to figure out how you want to communicate to peers...implementing the socket bits is super trivial, though
06:21 < BlueMatt> see https://github.com/TheBlueMatt/rust-lightning-bitcoinrpc/blob/master/src/net_manager.rs
06:25 < BlueMatt> ariard: oh shit, I had a comment about StaticOutputDescriptor written that I planned on posting pre-hike but failed :/
06:27 < BlueMatt> will respond to email-y things tomorrow-ish
06:27 < BlueMatt> (.nz time)
10:58 < stevenroose> github gave rust a new color :D
11:57 < ariard> shit I'm not a htlm developper :p
11:59 < ariard> BlueMatt: no worries busy learning other btc stuff, responded to your last comment, hope you got fun :)
13:08 < wumpus> stevenroose: apparently orange-red now! what did it use to be?
13:36 < stevenroose> wumpus: slightly lighter orange
13:36 < stevenroose> like brown-ish orange
13:36 < stevenroose> let me see if I have an old tab open
13:37 < stevenroose> wumpus: https://i.imgur.com/RnkS7Df.png
13:37 < stevenroose> how was my description? :D
13:38 < wumpus> (github's statistics used to be *really* off for bitcoin core, like, they labeled large part of the project as typescript, so I don't usually pay too much attention to that breakdown - but that's fixed now too)
13:38 < wumpus> ahh yes that's clear
14:12 <+andytoshi> mundane is a crypto library in rust that's designed to be hard to misuse. worth studying their api to see what ideas we can steal for rust-secp https://joshlf.com/post/2018/11/06/introducing-mundane/
14:15 <+andytoshi> oh wow, they somehow use symbol prefixng to avoid linker errors when compiling different versions of the underlying C lib (though they say this is a hack and that they need help, so maybe we can't steal that :))
14:39 < wumpus> there's quite a lot of rust crypto libraries appearing left and right, a few days ago heard about "orion" (https://github.com/brycx/orion)
14:41 <+andytoshi> maybe it'll induce the rustc developers to add constant time ops or zeroable memory to the stdlib..
14:46 < wumpus> yes... that was my concern as well
14:50 <+andytoshi> it's kinda frustrating, rust has a lot of great "high-assurance code" selling points and then it's basically useless for crypto because it bends over backward to let CPUs throw memory everywhere and reorder things
15:08 < wumpus> right, if the most risky parts of crypto code still end up being implemented in C that seems unfortunate
15:57 < wumpus> looks like there are RFCs for constant time and zeroable memory, but have been open for a long time: https://github.com/rust-lang/rfcs/issues/2533 https://github.com/rust-lang/rfcs/issues/1814
16:02 <+andytoshi> yeah. these things interact very poorly with both rust and LLVM :/
16:02 < wumpus> I guess the argument goes "this is only used for cryptography, which is only a very small subset of code, so doesn't belong in the standard library"
16:03 <+andytoshi> yeah they've given me crap like that to reject e.g. hex de/serialization in the standard library
16:03 <+andytoshi> "nobody uses sequences of bytes on a computer"
16:03 < wumpus> heh
16:03 < wumpus> I think it underestimates how *important* those little bits of code are compared to their volume
18:21 < stevenroose> going towards rust-secp native as a libsecp-C replacement? :)
19:24 <+andytoshi> heh, it would be nice if that were possible
19:56 <+andytoshi> published 0.1 of miniscript, which is the new name for the script descriptor crate
20:19 < stevenroose> andytoshi: could we perhaps keep the rust-[a-z-]+ convention for repos in rust-bitcoin :)
20:28 <+andytoshi> sure
20:28 <+andytoshi> renamed to rust-miniscript
20:29 <+andytoshi> it's a ways away from being a rust-bitcoin project because we still don't have a full spec for it upstream. sipa or I will present at SBC in january about it
21:46 < stevenroose> (secretly thinking about bitcoin_hashes)
21:47 < stevenroose> Actually I was wondering why bitcoin_hashes is outside of rust-bitcoin's core repo actually.  Seems to be used in every module in rust-bitcoin and virtually every bitcoin-related project
21:50 < stevenroose> dongcarl: how is the PSBT PR going? :D
22:13 <+andytoshi> stevenroose: bitcoin_hashes is in the rust-bitcoin org and afaik it is used nowhere because it's not on crates
22:19 -!- Topic for #rust-bitcoin: Rust-Bitcoin Libraries Discussion. Artisinal Bitcoin for the rest of us. Do NOT have private conversations on the FreeNode network! See https://github.com/rust-bitcoin If you created a repo, and I forgot to make you admin on it (stupid github defaults), ping BlueMatt
22:19 -!- Topic set by BlueMatt [~BlueMatt@unaffiliated/bluematt] [Sun Aug  5 09:06:05 2018]
22:19 [Users #rust-bitcoin]
22:19 [+andytoshi       ] [ dongcarl        ] [ grubles      ] [ mikethetike  ] [ sgeisler    ] [ warren ] 
22:19 [ _cryptosignal_me] [ dpc             ] [ gwillen      ] [ molz         ] [ simian_za   ] [ windsok] 
22:19 [ ariard          ] [ e4xit           ] [ harding      ] [ neonknight   ] [ stevenroose ] [ wraithm] 
22:19 [ belcher         ] [ elichai2        ] [ havik        ] [ nickler      ] [ TamasBlummer] [ wumpus ] 
22:19 [ Blackwolfsa     ] [ esotericnonsense] [ jamesob      ] [ savil        ] [ Terr        ] 
22:19 [ BlueMatt        ] [ gmaxwell        ] [ michaelsdunn1] [ scoobybejesus] [ Tralfaz     ] 
22:19 [ cjs77           ] [ gribble         ] [ midnightmagic] [ sdaftuar     ] [ treyzania   ] 
22:19 -!- Irssi: #rust-bitcoin: Total of 39 nicks [0 ops, 0 halfops, 1 voices, 38 normal]
22:19 -!- Channel #rust-bitcoin created Fri Mar  9 17:46:56 2018
22:20 -!- Irssi: Join to #rust-bitcoin was synced in 12 secs
22:38 < sgeisler> andytoshi: It is being used by rust-lightning-invoice ;)
23:03 < stevenroose> andytoshi: I was talking about "bitcoin hashes" in general, like current bitcoin::hash::Sha256dHash. Or is the idea to pull current bitcoin_hashes work into rust-bitcoin when it's done?
23:03 < stevenroose> btw, https://github.com/github/linguist/pull/4326
23:03 < stevenroose> Let's make a huge deal out of this :D
23:28 <+andytoshi> stevenroose: bitcoin_hashes is a conceptually very different crate from rust-bitcoin, and can be used independently ... e.g. i am considering making rust-secp depend on it
23:28 <+andytoshi> so that i can have infallible conversions from 32-byte hashes to rus-secp Messages