--- Log opened Thu Jun 20 00:00:48 2019
06:59 < dongcarl> fanquake: Yes, just gotta test a little more...
07:00 < dongcarl> Also, my substitute server is working at http://guix.carldong.io
07:00 < dongcarl> I'm also going to build the binary install tarball for my branch, so you guys don't have to do a long `guix pull` every time
07:01 < dongcarl> It is reproducible with `make guix-binary.x86_64-linux.tar.xz` in my guix branch... So you don't have to trust me
07:14 < cfields> dongcarl: woohoo!
07:14 < dongcarl> :-)
07:14 < dongcarl> soon^TM
07:15 < cfields> :)
08:55 -!- wumpus2 [~ircclient@pdpc/supporter/professional/wumpus] has joined #bitcoin-builds
08:58 -!- wumpus [~ircclient@pdpc/supporter/professional/wumpus] has quit [Disconnected by services]
08:58 -!- wumpus2 is now known as wumpus
09:49 < fanquake> :)
10:17 < wumpus> how does the kernel version end up influencing the build result?
10:17 < wumpus> that's kind of unexpected
10:17 < wumpus> (except if the different kernel version also comes with different headers, I suppose)
10:17 < dongcarl> wumpus: It didn't, the thing that ended up influencing the build result was the value of `$PWD`
10:18 < dongcarl> Yeah in our toolchain definition we pin linux headers to 4.19, which is the first LTS kernel that supports riscv
10:18 < dongcarl> We can push the linux header version back to earlier for other archs if need be. One line change.
10:19 < dongcarl> I think the `$PWD` might be an upstream gcc thing... Right now I just have the container bind $PWD to `/bitcoin` and build from there
10:20 < wumpus> $PWD makes sense, though I'd expect that to only influence the debug information, not the stripped executable?
10:20 < wumpus> oh
10:21 < wumpus> now that you say .. I remeber, yes there's definitely leaks of the build directory into the executable too, I've seen this in the past
10:21 < dongcarl> Yeah :-/
10:22 < dongcarl> We can figure it out later, binding seems to work fairly well, but I'm testing on different systems now
10:22 < wumpus> it was funny at the time as it was essentially state malware containing a /home/username
10:23 < dongcarl> Hahahahahaha
10:24 < dongcarl> Oy vey
10:26 < wumpus> I don't understand why, I mean for debug information it makes sense, you'd want it to be able to find the source code whereever it is
10:26 < wumpus> anyhow binding to a deterministic path will do it
10:27 < dongcarl> wumpus: Yeah, I'm adding it to my list of "nice things to figure out but the workaround is nice enough that it doesn't matter for now"
10:37 < MarcoFalke> review beg to build people: https://github.com/bitcoin/bitcoin/pull/16207/files ("test: stop generating lcov coverage when functional tests fail")
14:26 < dongcarl> MarcoFalke: don’t have keys to login to GitHub RN but LGTM
15:27 -!- Wendy40Jones [~Wendy40Jo@ns334669.ip-5-196-64.eu] has joined #bitcoin-builds
21:53 -!- midnightmagic [~midnightm@unaffiliated/midnightmagic] has joined #bitcoin-builds
--- Log closed Fri Jun 21 00:00:48 2019