| slangasek | doko: yes, that's exactly what it would mean, and what was discussed last week | 00:19 |
|---|---|---|
| slangasek | bdmurray: u-r-u rejected | 00:22 |
| doko | slangasek, that's not what I understood. I think we should discuss this again. the python-urllib3 itself should be considered in any case. | 07:20 |
| === fginther` is now known as fginther | ||
| === balloons is now known as Guest77290 | ||
| superm1 | cyphermox: oh that's too bad; it fixes that SHIM_DEBUG missing message. okay | 13:57 |
| cyphermox | superm1: next upload will need to go through signature process again, but that time should be faster | 14:11 |
| superm1 | cyphermox: okay. also should share this with you: https://github.com/rhinstaller/fwupdate/commit/91cb275c1bec0da0c61038892d980f1505e0ce3f | 14:12 |
| superm1 | that is supposed to fix that Dup error that has shown up occasionally | 14:12 |
| cyphermox | yep | 14:13 |
| superm1 | it's on the linux side of fwupdate though; so it shouldn't cause the efi binary to change | 14:13 |
| coreycb` | hello, can an archive admin please drop neutron-plugin-cisco neutron-plugin-metaplugin binaries? neutron no longer creates them. | 14:59 |
| cjwatson | coreycb`: removed from wily-proposed; that should let it migrate, and then the ones in wily will pop up on the NBS report for semi-auto removal | 15:07 |
| coreycb` | cjwatson, thanks | 15:08 |
| infinity | superm1: "shouldn't" and "won't" are two different things. But if the EFI binary is being built reproducibly, that would be awesome, so we can alter the userspace bits with confidence that we're not changing the EFI bits and can, thus, reuse the signed blob. | 15:38 |
| infinity | cyphermox: ^ | 15:38 |
| cyphermox | I don't know if it's reproducible, but I can check | 15:41 |
| cyphermox | that's not going to fix the sync though, because SHIM_DEBUG. | 15:41 |
| infinity | cyphermox: Oh, sure, I just meant in general. Like the commit cherrypick that superm1 pointed at, if that can be pulled in without altering the built EFI bits, then you can fudge things a bit for -signed and life goes on happily. | 15:42 |
| cyphermox | yeah | 15:43 |
| cyphermox | I'll dig into it later :) | 15:43 |
| infinity | cyphermox: Given the turnaround times we've sometimes seen on shim, if fwupdate has a slow-moving-efi/fast-moving-userspace thing going on right now, the only two options would seem to be either 100% reproducible binary output for the EFI bit, or splitting the package (or a third option of "we're always way behind", I guess). | 15:43 |
| cyphermox | superm1: could you file a bug about this? | 15:43 |
| cyphermox | infinity: turnaround is much worse for the first time ever signing the thing too. | 15:44 |
| flexiondotorg | cyphermox, Thanks for updating the ubuntu-mate-meta package yesterday :-) | 15:46 |
| superm1 | cyphermox: sure: https://bugs.launchpad.net/ubuntu/+source/fwupdate/+bug/1505741 | 16:19 |
| ubot93 | Launchpad bug 1505741 in fwupdate (Ubuntu) "Occasional errors with "dup: Bad file descriptor"" [Undecided,New] | 16:19 |
| cyphermox | superm1: thanks! | 16:20 |
| superm1 | infinity: the reason I said shouldn't is because I haven't confirmed it doesn't cause the EFI binary to change, but based on how the build system works it spits out userspace and EFI binary separately so I have confidence that it wouldn't. It just should be double checked. | 16:21 |
| === g4mby is now known as PaulW2U | ||
| cyphermox | superm1: I'm concerned whether some part of the build process doesn't mess things up -- namely, the self-signing that happens on the fwupdate EFI binary towards the end | 16:21 |
| infinity | superm1: Sure, and I suspect you're right about the EFI code not changing, but that doesn't guarantee the binary is identical, without occasionally taking steps to ensure that. | 16:22 |
| cyphermox | we could easily remove that though | 16:22 |
| superm1 | Yep | 16:22 |
| superm1 | cyphermox: The self signing you turned off in the last upload | 16:22 |
| cyphermox | ah | 16:22 |
| cyphermox | well, magic! | 16:22 |
| infinity | Heh. | 16:22 |
| cyphermox | superm1: I'll do a test in a bit rebuilding the binaries a couple of times and comparing the results | 16:23 |
| infinity | Anyhow, the whole "send a binary off to MS to sign and then package it up" workflow for shim (and now fwupdate) is messy as it is, so there's fudging either way, but we definitely want to make sure that if a user rebuilds the package, they're getting the same thing as the binary we ship in -signed, minus the sig. | 16:23 |
| superm1 | cool | 16:23 |
| * cyphermox -> lunch | 16:28 | |
| === coreycb` is now known as coreycb | ||
| robru | <queuebot> Unapproved: indicator-datetime <------------- this was published by mistake and the silo has been abandoned, please remove from queue | 20:09 |
| infinity | robru: The one in wily, from silo 46? | 20:49 |
| infinity | robru: Gone. | 20:50 |
| robru | infinity: yep, thanks | 20:52 |
| flexiondotorg | infinity, H-E-L-P! Please :-) | 21:36 |
| flexiondotorg | infinity, I've made a blunder. | 21:36 |
| flexiondotorg | infinity, Remember that artwork you merge for me recently. | 21:36 |
| flexiondotorg | Some of the images are the wrong bit depth and don't display at all, totally missing the point of being more visible for low vision users. | 21:37 |
| flexiondotorg | infinity, Please can you merge - https://code.launchpad.net/~ubuntu-mate-dev/debian-cd/ubuntu-mate-fix/+merge/274327 | 21:37 |
| flexiondotorg | cyphermox, Can you help? ^^^^^^^ | 22:10 |
| cyphermox | flexiondotorg: in a few hours | 22:21 |
| flexiondotorg | cyphermox, Thanks. I made a boo-boo :-/ | 22:25 |
| cyphermox | ok | 22:27 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!