19:03:23 <jonasschnelli> #startmeeting
19:03:23 <lightningbot> Meeting started Thu Aug 22 19:03:23 2019 UTC.  The chair is jonasschnelli. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:03:23 <lightningbot> Useful Commands: #action #agreed #help #info #idea #link #topic.
19:03:43 <jonasschnelli> There are no proposed topics
19:03:55 <jonasschnelli> (so far)
19:04:20 <meshcollider> #bitcoin-core-dev Meeting: wumpus sipa gmaxwell jonasschnelli morcos luke-jr sdaftuar jtimon cfields petertodd kanzure bluematt instagibbs phantomcircuit codeshark michagogo marcofalke paveljanik NicolasDorier jl2012 achow101 meshcollider jnewbery maaku fanquake promag provoostenator aj Chris_Stewart_5 dongcarl gwillen jamesob ken281221 ryanofsky gleb moneyball
19:04:23 <jonasschnelli> #topic high priority for review
19:04:27 <jonasschnelli> https://github.com/bitcoin/bitcoin/projects/8
19:04:30 <gleb> hi
19:04:33 <jonasschnelli> Anything to change?
19:04:49 <achow101> hi
19:05:33 <wumpus> test
19:05:38 <jonasschnelli> pong
19:05:43 <wumpus> wow, i'm visible now?
19:05:52 <meshcollider> wumpus: Yep
19:06:10 <wumpus> apparently I wasn't authenticated with freenode so no one was seeing my messages? this was really strange\
19:06:30 <wumpus> was already surprised why the meetingstart bot wasn't working
19:06:35 <achow101> how long has wumpus been shouting into the void?
19:06:37 <jonasschnelli> heh
19:06:40 <meshcollider> provoostenator has had similar issues before I think
19:06:42 <wumpus> about five minutes :)
19:07:01 <jeremyrubin> we miss anything good?
19:07:05 <jonasschnelli> omg... must have felt terrible
19:07:20 <wumpus> no, about the same jonasschnelli did :)
19:08:06 <harding> FYI,I saw "hi" or similar from the following people.  If you're not listed, you might be muted.  meshcollider jnewbery jonasschnelli provoostenator jeremyrubin ariard sipa dongcarl gleb achow101 (and, just now, wumpus).
19:08:08 <provoostenator> meshcollider: indeed, that's always fun
19:08:21 <wumpus> so anyhow, feel free to keep chairing the meeting
19:09:53 <meshcollider> So far no one has any topics
19:09:56 <jonasschnelli> Looks like non of the high prio PR's are ready to merge,...
19:10:05 <dongcarl> #proposedmeetingtopic upgrade to C++17
19:10:05 <jtimon> hi
19:10:21 <jonasschnelli> #topic upgrade to C++17 (dongcarl)
19:10:52 <dongcarl> I looked at the history a little, and there was a PR for upgrading to C++14
19:10:59 <wumpus> I think no one was prepared for that one
19:11:00 <jeremyrubin> What are the major new features?
19:11:14 <dongcarl> I believe that 17 has a lot of functionality that would simplify our codebase
19:11:26 <dongcarl> jeremyrubin: https://github.com/AnthonyCalandra/modern-cpp-features
19:11:34 <wumpus> would it replace any boost use?
19:11:47 <dongcarl> wumpus: Yes, I believe we rely on boost::optional
19:12:08 <wumpus> what would be minimum supported clang and gcc? which linux distros carry those? what about debian stable?
19:12:21 <wumpus> I think we have our own optional<>
19:12:27 <wumpus> but that's not really a big deal
19:12:48 <dongcarl> our own optional is just a wrapper around boost optional
19:12:54 <kanzure> hi
19:12:55 <dongcarl> Also c++17 has a filesystem module
19:13:05 <dongcarl> The minimum for ALL 17 is GCC8
19:13:22 <dongcarl> debian stable is on 8
19:13:27 <dongcarl> and RHEL dts is on 8
19:13:32 <achow101> are we going to just skip 14?
19:13:38 <wumpus> ubuntu 18.04 still has 7.4.0
19:13:45 <wumpus> ther's no way we're going to do this yet, imo, sorry
19:14:21 <dongcarl> Okay
19:14:27 <wumpus> achow101: 14 was not worth it, 17 probably is, but it's imo too early
19:14:32 <wumpus> but others might have other opinions
19:14:38 <dongcarl> Ubuntu has 8 packaged
19:14:49 <achow101> most of gcc 7 has 17
19:14:50 <wumpus> yes, but the default is 7.4.0
19:15:01 <achow101> err... gcc 7 has most of 14
19:15:06 <wumpus> I don't really want to cause people trouble for no good reason
19:15:09 <achow101> *17
19:15:22 <dongcarl> support matrix: https://gcc.gnu.org/projects/cxx-status.html#cxx17
19:15:23 <jeremyrubin> there doesn't seem to be too much new in 14+17
19:15:40 <jeremyrubin> the filesystem module seems compelling tho
19:15:47 <dongcarl> this is the only feature that requires GCC8: http://wg21.link/p0512r0
19:16:01 <jeremyrubin> But IIRC wumpus had some work that was replacing the fs module
19:16:07 <wumpus> has anyone tried replacing boost filesystem with it in our usage? anything missing?
19:16:08 <ryanofsky> wumpus, any people you are worried about in particular? i'd expect most non-developers to just use binaries and not be affected
19:16:22 <sipa> for a long time we had C++11-minus-thread_local as guideline; perhaps there is something similar for C++17 that would bring the compatibility up?
19:16:26 <wumpus> ryanofsky: I'd like bitcoind to be easily buildable on a wide range of systems such as VPSes and such
19:16:49 <wumpus> many people build bitcoin from source
19:17:08 <sipa> i'd like c++17, but not until it's easily supportable on all reasonable systems
19:17:16 <ryanofsky> so we are worried about the type of person who knows how to build bitcoin from source, but not how to install an updated compiler
19:17:16 <dongcarl> GCC7 has everything but one obscure C++17 feature: https://gcc.gnu.org/projects/cxx-status.html#cxx17
19:17:26 <dongcarl> What systems are we concerned about?
19:17:41 <wumpus> please, installing an updated compiler can be nasty on some platforms
19:18:03 <wumpus> gcc7 as requirement might be ok, but 8 is not
19:18:09 <ryanofsky> wumpus, i guess i'm not aware...
19:18:38 <dongcarl> I'm sorry I misspoke
19:18:49 <sipa> also, there is a historical case of a miner not wanting to update because he couldn't get a c++11 compatible compiler or something
19:18:57 <wumpus> I think we should have this discussion on github, not in the meeting
19:19:25 <MarcoFalke> I'd say that installing a package (compiler or anything) not through a package manager should not be encouraged by us
19:19:26 <dongcarl> Okay
19:19:38 <ryanofsky> yeah, it'd be good to know specifics of who would be affected
19:19:40 <jnewbery> why is this off-topic for a meeting?
19:19:43 <meshcollider> So C++17-class_template_argument_ddeduction might be ok then
19:19:43 <wumpus> MarcoFalke:yes it sounds strange to me
19:19:51 <wumpus> jnewbery: it's not off topic
19:20:13 <wumpus> but I think this requires a fair amount of research, to see what is affected and what not
19:20:24 <wumpus> and als o*why* we want this
19:20:36 <wumpus> and maybe someone actualy implementing the fs stuff using std filesystem and seeing what comes up
19:20:57 <wumpus> it seem a bit random to just propose 'oh let's bump the requirement to c++17'
19:21:07 <MarcoFalke> There was an issue on C++14 and it was rejected because centos and similar didn't ship a C++14 compiler. Unless this changes, we don't even need to chat about C++17
19:21:10 <wumpus> but that might be just me
19:21:40 <sipa> well when we eventually want to switch to c++17 we will want to discuss it in the meeting
19:21:46 <dongcarl> I'm sorry, I might be mis-remembering, but in the previous C++14 PR, someone requested for things like this to be discussed in a meeting
19:21:47 <jnewbery> I think it's useful to solicit opinions in a meeting
19:21:50 <dongcarl> So that's what I did
19:21:55 <sipa> but it sounds like we may not have enough information right now
19:21:59 <wumpus> dongcarl: sure
19:22:13 <dongcarl> I'll open an issue, we'll discuss there
19:22:19 <jonasschnelli> ack
19:22:22 <wumpus> ack, thanks
19:22:56 <jonasschnelli> I guess it needs a rational. Benefits / risks / user-costs
19:23:14 <wumpus> right
19:23:16 <jonatack> hi
19:23:41 <meshcollider> Centos still has 4.8.5 from what I can see
19:23:57 <MarcoFalke> jup
19:24:37 <MarcoFalke> So first we'd need to wait for centos 8 to come out and then give users plenty of time to upgrade
19:25:12 <wumpus> it'd be too late for 0.19 anyhow, but probably not 0.20 (in half a year) then either
19:26:02 <MarcoFalke> https://wiki.centos.org/About/Building_8
19:26:22 <MarcoFalke> apparently they have an rc out
19:26:31 <wumpus> a reminder that 0.19 release date is getting pretty close, soft string freeze is 2019-09-01, feature freeze 2019-09-15, that's less than a month
19:26:51 <sipa> thanks
19:27:07 <wumpus> we probably should start prioritizing features for review that need to make it in for 0.19
19:27:15 <wumpus> or at least, would be nice to make it in
19:27:17 <MarcoFalke> quick. Merge all the things!111!!!
19:27:31 <wumpus> heh
19:29:28 <wumpus> ok, any other topics?
19:30:27 <wumpus> #endmeeting
19:30:33 <jonasschnelli> #endmeeting