**** BEGIN LOGGING AT Wed Sep 29 02:59:56 2021
Sep 29 07:45:53 <mckoan> good morning
Sep 29 09:10:14 <mort> I feel like there should be an env var called GO which refers to aarch64-poky-linux-go in the sdk
Sep 29 12:23:04 <marex> RP: I might need to inherit meta/classes/reproducible_build.bbclass:BUILD_REPRODUCIBLE_BINARIES ??= '1'
Sep 29 12:23:08 <marex> which I do not
Sep 29 12:24:39 <RP> marex: fair enough. I think we added testing for it in dunfell but not by default. In master it is the default in poky
Sep 29 12:25:04 <marex> RP: are you aware of some more ... standard ... python performance benchmark ?
Sep 29 12:25:16 <marex> s@python@& interpreter@
Sep 29 12:25:23 <marex> I am not that deep into python myself
Sep 29 12:46:45 <RP> marex: I know pgo training workloads have a big influence
Sep 29 12:47:26 <marex> RP: but debian also has PGO enabled
Sep 29 12:47:47 <marex> maybe they train it on something else ?
Sep 29 12:48:23 <RP> marex: that is what I'm wondering
Sep 29 12:48:27 <marex> I wonder if there is some comprehensive benchmark, so I could rule out that this one particular piece of python code is an outlier
Sep 29 12:50:13 <RP> marex: https://github.com/python/pyperformance looks promising?
Sep 29 12:54:00 <marex> RP: well, lemme build an image with pip and try it
Sep 29 16:45:19 <khem> marex: pgo is subjective, we use qemu to run in cross compile usermode and it might be noot resulting in accurate latencies and also check if we are curtailing the pgo run to save on buildtime, because you can easily spend an hr just building python with pgo on some arches
Sep 29 17:38:55 <marex> khem: hmmm, that is a valid point, debian is building native
**** ENDING LOGGING AT Thu Sep 30 02:59:56 2021