--- Log opened Fri Oct 01 00:00:23 2021
00:11 < laanwj> I'm somewhat disappointed non one tested the 22.0 RCs on PowerPC at all, apparently?
00:11 < laanwj> re: https://github.com/bitcoin/bitcoin/issues/23111
00:12 -!- jonasschnelli [~jonasschn@static.239.36.216.95.clients.your-server.de] has quit [Ping timeout: 252 seconds]
00:12 < laanwj> if they did, can we be sure every powerpc system even uses the same dynamic linker path?
00:13 < laanwj> i remember ARM having two conventions (depending on the softfloat/hardfloat ABI)
00:13 -!- jonasschnelli [~jonasschn@2a01:4f9:2a:2510::2] has joined #bitcoin-core-builds
00:15 < laanwj> full static linking with musl wen
00:15 < fanquake> tomorrow
00:15 < laanwj> !!!
00:35 < fanquake> right after LTO
00:39 < laanwj> yea full static build + lto is an interesting combination
00:39 < laanwj> everything can be fused and dissolved into everything else
00:42 < laanwj> though last time i tried lto with bitcoin core it took hours and eventually ran out of memory :-)
00:42 < laanwj> but that's a long time ago, compilers are better now
00:42 < fanquake> then just make everything constexpr, and the compiler will produce you a blob that doesn't do much at all
00:43 < laanwj> yess
00:43 < fanquake> heh. I tried a few LTO builds today, linking qt can certainly take a few minutes
00:44 < laanwj> right, pretty much all work moves to the linking phase
01:02 < laanwj> going to try it on RISC-V 
01:02 < fanquake> I'll just go ahead and add a commit to that PR adding that flag to the Guix build, to facilitate testing those as well
01:05 < laanwj> I think the information printed at the end of configure is getting out of hand
01:05 < laanwj> it's good to print a summary of which features are enabled, but I think we're adding some internals that aren't useful to most developers/builders
01:06 < laanwj> (I mean, what you want to avoid is that it's just like the endless configure-scroll itself)
01:06 < laanwj> fanquake: sgtm
01:07 < laanwj> a great idea even, makes testing it in cross-compilation context easier
01:16 -!- ryanofsky [~russ@jumpy.yanofsky.org] has quit [Ping timeout: 250 seconds]
01:16 -!- ryanofsky [~russ@jumpy.yanofsky.org] has joined #bitcoin-core-builds
01:23 < fanquake> laanwj: fair, bit of a trade-off there in regards to what to display. Certainly info that most might not find too interesting
01:23 < fanquake> I pushed that Guix commit up, running a build now too
01:30 < fanquake> Looks like ar / ranlib in Guix wont work with LTO our of the box
01:37 < laanwj> fanquake: yes, it's definitely a trade-off, i think the sweet spot is that it fits on one screen and mostly shows yes/no, information that can not be included can be logged as part of the normal output in-line
01:38 < laanwj> ar/ranlib? that's interesting, it's basically a dumb archival tool :)
01:42 -!- kanzure [~kanzure@user/kanzure] has quit [Ping timeout: 252 seconds]
01:42 -!- greypw2 [~greypw2@grey.pw] has quit [Ping timeout: 252 seconds]
01:42 -!- greypw2 [~greypw2@grey.pw] has joined #bitcoin-core-builds
01:42 < laanwj> i don't think "ar" is supposed to do anything special in lto, all the linking happens when making the binary, but it does store some metadata like symbol tables for faster lookup maybe that fails
01:42 -!- glozow [sid453516@user/glozow] has quit [Ping timeout: 252 seconds]
01:43 -!- kanzure [~kanzure@user/kanzure] has joined #bitcoin-core-builds
01:43 < fanquake> See here for a couple failed builds: https://gist.github.com/fanquake/9c3d194614955a3c2e3e969c5a7c4812
01:44 < fanquake> ar "plugin needed to handle lto object"
01:44 < laanwj> i see
01:45 < laanwj> i do remember the hassle with ld 'gold' plugin in the beginning when lto was new, i didn't know ar had a similar mechanism
01:45 < fanquake> Note that I haven't seen any issues building outside guix yet, across macos, linux etc, so likely some things we need to setup / fix inside the Guix env
01:45 < laanwj> probably the toolchains are built without lto support entirely
01:57 -!- glozow [sid453516@user/glozow] has joined #bitcoin-core-builds
02:12 -!- kanzure_ [~kanzure@user/kanzure] has joined #bitcoin-core-builds
02:14 -!- kanzure [~kanzure@user/kanzure] has quit [Ping timeout: 252 seconds]
04:22 < hebasto> 07:15:25 <laanwj> full static linking with musl wen -- Already possible in Alpine, but I assume you mean non-Alpine OS, right?
04:23 -!- CoinForensics [CoinForens@gateway/vpn/protonvpn/coinforensics] has joined #bitcoin-core-builds
04:23 < hebasto> fwiw, https://gist.github.com/hebasto/a7f08e44ddb79cc4cde2762cf9ec5688
04:25 < laanwj> hebasto: i know it's possible, but i mean for the release binaries
04:26 < hebasto> that would be great
04:26 < laanwj> right-it would be great to not depend on a specific dynamic linker / libc, there have been various issues around this, luckily minor ones but still
05:35 -!- jtrag [~jtrag@user/jtrag] has joined #bitcoin-core-builds
06:53 < laanwj> "lto1: internal compiler error: Segmentation fault" interesting, but doesn't seem reproducible, might have been an intermittent issue with the board
06:53 < laanwj> good ways to stress test your CPU and I/O: bitcoin functional tests, compilation with lto :-)
06:55 < sipa> could it be OOM?
06:57 < laanwj> it could be! i didn't think so because the board has 16G of memory, but that's "only" 4GB per thread, and i don't have swap enabled
07:19 -!- jtrag [~jtrag@user/jtrag] has quit [Quit: <-----  is PODAK (Passed out drunk at keyboard), and he has somehow managed to quit/disconnect...]
07:19 < laanwj> weird to see -Wmaybe-uninitialized warnings during linking (though it makes sense)
07:55 -!- CoinForensics [CoinForens@gateway/vpn/protonvpn/coinforensics] has quit [Ping timeout: 252 seconds]
08:35 < fanquake> laanwj: yes I saw some in the fuzz tests as well
08:59 -!- cfields [~cfields@user/cfields] has quit [Ping timeout: 252 seconds]
09:03 -!- cfields [~cfields@user/cfields] has joined #bitcoin-core-builds
09:19 < MarcoFalke> Yall are probably using gcc?
09:22 < sipa> you guys use a compiler?
09:48 < laanwj> yes, gcc here
09:48 < laanwj> sipa: i'm not capable of compiling c++ without compiler yet
09:49 < sipa> laanwj: sounds like a bootstrapping problem ;)
09:49 < laanwj> oh no not another one :)
09:49 < sipa> :p
11:25 -!- gribble [~gribble@bitcoin/bot/gribble] has joined #bitcoin-core-builds
12:54 -!- jtrag [~jtrag@user/jtrag] has joined #bitcoin-core-builds
13:03 -!- kanzure_ is now known as kanzure
13:30 -!- jamesob [sid180710@id-180710.helmsley.irccloud.com] has quit [Quit: Connection closed for inactivity]
16:21 -!- belcher [~belcher@user/belcher] has quit [Ping timeout: 252 seconds]
16:33 -!- belcher [~belcher@user/belcher] has joined #bitcoin-core-builds
17:56 -!- cfields_ [~cfields@user/cfields] has joined #bitcoin-core-builds
17:57 -!- cfields [~cfields@user/cfields] has quit [Ping timeout: 252 seconds]
21:13 -!- sdaftuar1 [~sdaftuar@user/sdaftuar] has quit [Ping timeout: 260 seconds]
21:14 -!- sdaftuar1 [~sdaftuar@user/sdaftuar] has joined #bitcoin-core-builds
--- Log closed Sat Oct 02 00:00:24 2021