[00:37] <xnox> Odd_Bloke:  cjwatson: isn't it the first one in /var/lib/dpkg/arch ?!
[00:37] <xnox> but i think that is internal / private, not public api
[02:46] <mwhudson> ENOENT
[02:49] <mwhudson> dpkg seems to set it at build time
[08:52] <didrocks> cpaelzer: thanks for the libos-info patch (and yeah, the readme needs some update it seems :p)
[09:12] <cpaelzer> didrocks: yw
[09:12] <cpaelzer> for once I tried to follow the "how to contribute" and then it is outdated :-)
[09:12] <cpaelzer> didrocks: did you spot any issues in the MR ?
[09:13] <cpaelzer> and if not maybe say "LGTM" there as you also are listed as Reviewed-by on the last Ubuntu changes
[09:15] <didrocks> cpaelzer: I’ll have a deeper look this morning, but at first glance, I didn’t spot anything. Once done, yes, I’ll answer on gitlab
[09:20] <cpaelzer> thank you didrocks
[11:50] <tkamppeter> Some core dev around here? Could you have a look at bug 1862926? It updates SANE to make the scanning in nearly all modern multi-function printers work and hopefully also stops users from mucking around with HPLIP (bug 1766020).
[14:23] <LocutusOfBorg> tkamppeter, .
[14:23] <tkamppeter> LocutusOfBorg, ? ?
[14:24] <LocutusOfBorg> 16mb of debdiff... meh
[14:24] <LocutusOfBorg> can't we help uploading in debian experimental and then go for ubuntu?
[14:24] <tkamppeter> LocutusOfBorg, the huge debdiff is due to the upstream code included.
[14:25] <tkamppeter> LocutusOfBorg, an new upstream version always causes a big debdiff, especially if it is two feature releases later.
[14:29] <LocutusOfBorg> yes, but I would prefer to upload on debian expeirmental
[14:29] <LocutusOfBorg> and then sync
[14:33] <tkamppeter> LocutusOfBorg, could you do that? We need to get it done before FF.
[15:29] <cpaelzer> cjwatson: hiho - a binfmt question, qemu's style to use this is very outdated
[15:30] <cjwatson> cpaelzer: I'm pretty sure there's a Debian bug about this
[15:30] <cpaelzer> yes
[15:30] <cpaelzer> 866756
[15:30] <cpaelzer> but I have a detail question abotu it for you as binfmt expert
[15:30] <cjwatson> Sure
[15:31] <cpaelzer> we had a problem e.g. in WSL containers that the (old style) direct calls to "update-binfmts ... --install" failed there
[15:31] <cpaelzer> the fix in the past was to have a better contianer detection
[15:31] <cjwatson> --import shouldn't have that problem
[15:31] <cpaelzer> but reading through the bug above I was wondering what binfmt triggers will do
[15:32] <cjwatson> I mean, by all means try it out, but it isn't supposed to fail if the equivalent of --install fails
[15:33] <cpaelzer> so it would try and gracefulyl go on
[15:33] <cjwatson> more or less on the grounds that the information is still on disk and can be replayed, yes
[15:33] <cpaelzer> cjwatson: do you have a package in mind that already uses the --import + trigger correctly?
[15:34] <cjwatson> please don't say trigger here - binfmt-support probably should support actual triggers, but it doesn't yet (https://bugs.debian.org/945019)
[15:36] <cjwatson> I don't know of anything using --unimport correctly (which isn't to say they don't exist); quite a few use --import (e.g. pythonX.Y, llvm-8-runtime).  But really if you follow the detailed instructions in /usr/share/doc/binfmt-support/README.Debian you should be fine
[15:36] <cpaelzer> thanks cjwatson
[15:37] <cjwatson> The only qemu quirks AFAIK are that it has arch-specific binfmts, and that it needs to be extra-careful to avoid accidentally registering a binfmt for the native architecture (which is a great way to instantly lock up your system)
[15:40] <cpaelzer> the omit logic didn't change for years and I'd not touch it on that change
[15:49] <cjwatson> I suppose you can keep on computing that at run-time and just import the right set of binfmt files
[15:49] <cjwatson> I'm not sure that will work well though
[15:50] <cjwatson> cpaelzer: you really want to ship the right set of binfmt files at build time rather than at run-time
[15:50] <cjwatson> cpaelzer: doing it at build time means they can all go in /usr/share/binfmts/, which means that if binfmt-support is installed after qemu then it can run 'update-binfmts --import' and catch up
[15:50] <cjwatson> cpaelzer: this is not going to work well if you keep the arch-specific handling entirely in the postinst
[17:07] <tkamppeter> LocutusOfBorg, reported SANE update request to Debian as Debian bug 951213.
[17:10] <LocutusOfBorg> tkamppeter, will have a look tomorrow hopefully
[17:13] <tkamppeter> LocutusOfBorg, are you Debian developer?
[18:15] <LocutusOfBorg> tkamppeter, yes I am