#archlinux-ports | Logs for 2017-06-28
Back
[01:52:44] -!- Faalagorn1 has joined #archlinux-ports
[01:55:23] -!- Faalagorn has quit [Ping timeout: 246 seconds]
[02:22:17] -!- fhdrin has joined #archlinux-ports
[02:46:07] -!- eschwartz has quit [Remote host closed the connection]
[02:53:01] -!- eschwartz has joined #archlinux-ports
[02:53:06] -!- fhdrin has quit [Quit: Leaving]
[03:22:20] -!- eschwartz has quit [Ping timeout: 268 seconds]
[03:25:43] -!- eschwartz has joined #archlinux-ports
[03:47:18] -!- eschwartz has quit [Ping timeout: 260 seconds]
[04:04:01] <guys> Well, sure. linux-hardened isn't currently built even though i686 is supported.
[04:04:29] <guys> IIRC there was something said about how lots of linux-hardened stuff has no meaning on i686
[04:06:33] <fsckd> perhaps ask strcat if he's interested in getting it working on i686.
[04:08:05] <fsckd> they may be willing to help or at least offer some advice or tips.
[05:04:41] -!- isacdaavid has joined #archlinux-ports
[05:07:36] -!- fhdrin has joined #archlinux-ports
[05:31:16] -!- fhdrin has quit [Ping timeout: 276 seconds]
[06:28:48] -!- deep42thought has joined #archlinux-ports
[06:44:05] -!- titus_livius has joined #archlinux-ports
[06:59:15] -!- deep42thought has quit [Remote host closed the connection]
[07:20:28] -!- eschwartz has joined #archlinux-ports
[07:43:05] -!- deep42thought has joined #archlinux-ports
[07:44:01] -!- Faalagorn1 has quit [Remote host closed the connection]
[07:47:56] -!- Faalagorn has joined #archlinux-ports
[08:00:15] -!- Faalagorn has quit [Remote host closed the connection]
[08:04:35] -!- isacdaavid has quit [Quit: isacdaavid]
[08:08:08] -!- eschwartz has quit [Remote host closed the connection]
[08:24:23] -!- deep42thought has quit [Quit: Leaving.]
[10:01:46] -!- dmakeyev has quit [Ping timeout: 246 seconds]
[11:01:54] -!- deep42thought has joined #archlinux-ports
[11:04:35] -!- Guest29858 has quit [Changing host]
[11:04:35] -!- Guest29858 has joined #archlinux-ports
[11:05:39] Guest29858 is now known as jelle
[13:11:16] -!- dmakeyev has joined #archlinux-ports
[13:42:02] -!- eschwartz has joined #archlinux-ports
[14:30:28] -!- tyzoid has quit [Ping timeout: 260 seconds]
[14:33:44] -!- tyzoid has joined #archlinux-ports
[14:37:05] -!- tyzoid has quit [Client Quit]
[14:37:36] -!- tyzoid has joined #archlinux-ports
[15:53:05] -!- dmakeyev has quit [Ping timeout: 255 seconds]
[16:30:28] -!- deep42thought has quit [Quit: Leaving.]
[16:43:59] -!- Faalagorn has joined #archlinux-ports
[17:11:36] -!- eschwartz has quit [Remote host closed the connection]
[17:14:03] -!- eschwartz has joined #archlinux-ports
[17:31:16] -!- Faalagorn has quit [Remote host closed the connection]
[17:32:02] -!- Faalagorn has joined #archlinux-ports
[18:07:46] <brtln> i686 folks, can I ask you to test mkinitcpio-busybox from testing?
[18:08:54] <tyzoid> Is there a new version from last week?
[18:08:58] <tyzoid> brtln: ^
[18:09:22] <eschwartz> brtln: Well, as soon as I reboot I will have checked it?
[18:09:38] <brtln> 2017-06-26 06:36
[18:09:42] <brtln> I guess it doesn't count as last week
[18:09:42] <tyzoid> ah
[18:09:50] <brtln> eschwartz: mkay
[18:10:12] <eschwartz> I did hold off on updating for a few days since I was trying to figure out why Thunderbird segfaults in the IRC tab
[18:10:29] <brtln> TIL someone uses thunderbird for irc
[18:10:42] <brtln> tyzoid: you should consider becoming a tester for the remaining time of i686 being tier1
[18:11:07] <eschwartz> It appears to be the fault of gcc 7 somehow
[18:11:32] <brtln> sounds like potential bug assigned to me
[18:11:33] * brtln covers his ears
[18:11:45] <tyzoid> brtln: Didn't we have this issue a while ago?
[18:12:02] <tyzoid> I needed to hold back the glibc version to get the system to boot
[18:12:12] <eschwartz> I rebuilt thunderbird 52.1.1 (libevent rebuild) and it started crashing too, so I compiled aur/gcc6 and rebuilt thunderbird again and that's what I am using now.
[18:12:28] <brtln> tyzoid: well, it's different one
[18:12:51] <brtln> tyzoid: I'm going to keep mkinitcpio-busybox built statically though for now… feel free to check if dynamic rebuild works
[18:12:57] <eschwartz> currently compiling thunderbird 52.2.1 to update+check that out
[18:13:31] <brtln> eschwartz: yeah, I'm considering adding gcc ${ver--} to [extra]
[18:13:46] <eschwartz> brtln: I should probably report that to Mozilla though, I assume.
[18:13:51] <brtln> sounds like something useful in general
[18:14:01] <eschwartz> we do have gcc5
[18:14:19] <brtln> don't mention it with me present
[18:14:49] <brtln> that's something that should die with anything depending on it
[18:14:57] <eschwartz> sure.
[18:15:23] <eschwartz> That's why I built gcc6 from the AUR for my test
[18:15:29] <tyzoid> brtln: I'm building up a new arch32 iso out of mainline/testing
[18:15:35] <tyzoid> If that boots, everything should be good
[18:15:48] <brtln> tyzoid: thx
[18:16:24] <tyzoid> and brtln: I'm sure you're familiar with the automated testing we're planning on starting
[18:16:42] <brtln> I don't know what you decided to do
[18:16:44] <tyzoid> not sure if me becoming an arch tester would impact that/allow us to use that infrastructure in mainlin too
[18:16:45] <brtln> openqa?
[18:17:01] <tyzoid> brtln: still in planning stages: https://bbs.archlinux32.org
[18:17:03] <phrik> Title: Discussion of possible means of testing / Testing / Arch Linux 32 Forums (at bbs.archlinux32.org)
[18:17:40] <tyzoid> the current idea is to script test cases inside containers
[18:17:56] <brtln> how does that test initcpio and kernel?
[18:18:10] <tyzoid> automated iso build testing
[18:18:12] <brtln> sounds pretty weak
[18:18:36] <tyzoid> We'd use a vm for more complex things, but the containers were for speed of testing
[18:18:48] <tyzoid> there're 2000+ packages that we'd like to have at least some tests for
[18:18:58] <tyzoid> brtln: The approach we're going with is breadth first, not depth first
[18:19:05] <tyzoid> We'd like to make a system that could do both
[18:19:14] <tyzoid> but we need *some* tests for things
[18:20:02] <tyzoid> hopefully that makes some sense
[18:20:30] <tyzoid> I don't know the extent at which some of these packages are tested, but afaik, it's mostly just the testers using them as builds come in.
[18:20:49] <tyzoid> brtln: Perhaps you could confirm/deny that?
[18:21:01] <tyzoid> but that was my impression after talking to some of the devs in #archlinux
[18:21:08] <eschwartz> more or less, yeah
[18:21:15] <brtln> yes, there is no automated testing beyond check()
[18:21:53] <brtln> please get back to me when you settle on something… but still openqa sounds like something that covers a lot of smoke tests already
[18:22:55] <brtln> really wish I had time to handle any kind of CI myself but I don't so I guess we will be happy to re-use anything you do
[18:23:18] <tyzoid> brtln: Sounds good, and I'll definately look at openqa
[18:24:26] <brtln> auto build infra is something I'd need to do over Allan's dead body so it's a no go, but for example continous rebuilds of repo to see if everything is fine would also be something we could adopt
[18:25:05] <brtln> there was so arch-ci instance for a while but I don't know who did that and why he stopped
[18:25:11] <brtln> if you're here, say hi
[18:25:27] <tyzoid> I'm not sure they'll be in arch-ports
[18:27:00] * brtln sighs
[18:27:02] <brtln> what can I say
[18:27:59] <tyzoid> brtln: Regardless, I've got some free time tonight, I'll see if I can't get some tests of openqa going for the archiso+arch32iso
[18:28:36] <tyzoid> rewbycraft has an openstack instance, so we might be able to leverage that for doing the tests
[18:29:49] <tyzoid> brtln: Do you have an account on the forum?
[18:30:24] <brtln> no, I don't plan to have one
[18:30:37] <tyzoid> fair enough
[18:31:04] <brtln> I hoped you will discuss everything on arch-ports but apparently not :P
[18:31:07] <brtln> as in, mailing list
[18:33:10] -!- kerberizer has quit [Ping timeout: 240 seconds]
[18:36:30] <tyzoid> brtln: mkinitcpio-busybox 1.26.1-1 works with linux 4.11.7-1 and glibc 2.25-5
[18:36:33] <tyzoid> on i686
[18:37:40] -!- kerberizer has joined #archlinux-ports
[18:38:17] <tyzoid> brtln: We do most of our discussions here, but we use the forums as a publicly accessible archive/summary
[18:38:32] <tyzoid> pretty much everything there has already been discussed here
[18:38:45] <tyzoid> and it's more documentative to help us keep track of things
[18:38:49] <tyzoid> email doesn't really do that
[18:40:05] <brtln> your call, I don't really help you so I can't decide about where you do it
[18:40:41] <tyzoid> well most of us monitor all three places, so if you ever want us for something, you know where to holler
[18:41:04] <brtln> but forums are inconvienent as it's another place I'd have to watch
[18:41:09] <brtln> I don't keep up with backlogs either
[18:41:21] <tyzoid> brtln: The forums can send you an email digest
[18:41:43] <tyzoid> we use the same software as the main arch forum
[18:41:54] <tyzoid> so we have the same functions as over in mainline
[18:43:35] <tyzoid> also brtln: I assume you saw my +1 for mkinitcpio-busybox?
[18:43:59] <brtln> yes, moved it to [core] 10s ago
[18:45:26] <tyzoid> sweet
[18:50:31] <tyzoid> hey brtln: We're thinking forward to when we need to start migrating people off of the mainline repos and onto our own repos
[18:50:49] <tyzoid> The problem is that we have our own keyring, archlinux32-keyring in our repos that we sign our packages with
[18:51:18] <tyzoid> I'm wondering what a good way to be able to get users these keys without having to blindly trust our package
[18:58:21] <brtln> well there is no good solution I guess
[18:58:39] <brtln> we can do it cat and mouse way of verifying you like master key holders do in mainline
[18:59:18] <brtln> then you all sign your package, put somewhere besides the repo, and I sign it too after verifying that you really did sign it
[18:59:28] <brtln> this way pacman won't complain
[18:59:30] -!- kerberizer has quit [Ping timeout: 240 seconds]
[19:00:02] <brtln> (because it will verify my signature against archlinux-keyring)
[19:00:33] <tyzoid> brtln: Actually, that should work. If we have an arch mainline contributor sign the package, then we should be set
[19:04:39] -!- kerberizer has joined #archlinux-ports
[19:08:10] <jelle> do you want to use the opensuse openqa?
[19:08:35] <jelle> tyzoid: ^
[19:15:13] <rewbycraft> That's some neat software
[19:15:30] * jelle has mixed experiences with it
[19:15:34] <jelle> a) it's perl
[19:15:41] <jelle> b) it's opensuse ware
[19:15:41] <rewbycraft> And it sounds like a lot of moving parts
[19:15:58] <jelle> c) never seen it used outside of opensuse
[19:16:31] <rewbycraft> Seems like some non-trivial infra is needed for it
[19:16:44] <jelle> also, haven't seen documentation at all :P
[19:16:59] <rewbycraft> open.qa/documentation
[19:17:32] <jelle> anyway I'm kinda against separate tests which every distro has to implement
[19:18:25] <jelle> debian btw has something called autopkgtests
[19:19:57] <jelle> been some time though since I checked it all out :)
[19:20:23] <rewbycraft> I dislike most of the stuff around packaging that debian does
[19:20:44] <rewbycraft> Making .debs of anything is a painful exercise in the digital equiv of paperwork
[19:20:47] <jelle> ideally I would see something like ```sudo extra-$arch-build which after building, installs pkg (already does) and then runs tests (but what)
[19:21:04] <jelle> rewbycraft: totally beside the point ;-)
[19:21:22] <rewbycraft> It's just the main reason I avoid looking at debian for anything packaging related
[19:21:46] <tyzoid> jelle: This is one potential solution
[19:22:16] <tyzoid> I'd like to get my feet wet with it first, so that we're not all just sitting here talking about what we could do :)
[19:22:22] <tyzoid> I'd like to get testing implemented asap
[19:22:32] <jelle> k
[19:22:34] <tyzoid> so if you've got better ideas...
[19:22:57] <jelle> I have very little time this month :)
[19:23:02] <tyzoid> our original solution was going to a tangle of bash scripts anyway before it got formalized
[19:23:11] <jelle> k
[19:27:33] -!- deep42thought has joined #archlinux-ports
[19:37:46] <deep42thought> tyzoid: Regarding the transition, instead of one mainline packager signing our archlinux32-keyring package, we could also let three of them sign our master keys
[19:38:04] <tyzoid> I think that would be less likely to happen
[19:38:08] <tyzoid> but still possible
[19:38:12] <tyzoid> wb deep42thought
[19:38:16] <deep42thought> Hi
[19:38:39] <deep42thought> well, don't we already have two trusted users on our team?
[19:54:10] <tyzoid> if we do, then we can have them sign that package
[20:05:43] <brtln> I don't see what it brings
[20:06:01] <brtln> it won't make your keys trusted in pacman eyes
[20:06:20] <deep42thought> pacman trust every thing which is signed with a "trusted" key
[20:06:21] <tyzoid> brtln: If we have a trusted user sign the archlinux32-keyring, then the keyring is
[20:06:30] <deep42thought> and trusted is everything which is signed by 3 trusted keys
[20:06:42] <tyzoid> so the user installs the archlinux32-keyring and then can install our packages
[20:06:45] <tyzoid> that's how I'm thinking
[20:06:48] <deep42thought> or what is signed by the private key
[20:06:55] <deep42thought> s/private/local/
[20:07:30] <deep42thought> or we get our >=3 master keys signed by >=3 trused users and would not need archlinux32-keyring at all
[20:07:37] <deep42thought> at least not if archlinux-keyring is installed
[20:08:16] <brtln> there is some wat density I can't handle
[20:08:32] <eschwartz> wat
[20:08:33] <brtln> looks you know something about how our keyring knows that I don't
[20:08:47] <brtln> you need master key holders to sign your keys if you want to look it this way
[20:08:55] <deep42thought> the keyring is just the usual gpg trust model - as far as I understood
[20:08:58] <brtln> and you always need a keyring unless you want annoying pacman questions
[20:09:04] <eschwartz> archlinux32-keyring signed by valid key == archlinux32-keyring can install its own master key with ultimate trust
[20:09:44] <brtln> what I propose all the time is that someone from arch packages and signs the first 32bit keyring for you
[20:09:51] <brtln> just for transition purposes
[20:10:37] <deep42thought> that's possible, too, but it would require us to have the same key in our keyring, because otherwise archlinux32-keyring won't be valid signed by its own measures ...
[20:10:43] <brtln> I don't see what you really want achieve by having trusted users sign your keys, but it's not going to work like you described so far
[20:11:30] <brtln> you can just name the package archlinux32-keyring-migration or whatever
[20:11:40] <brtln> you won't need it after user installs it
[20:11:46] <deep42thought> ah, good point
[20:11:50] <tyzoid> eschwartz: That's what I'm trying to say
[20:11:51] <brtln> archlinux32-keyring from your repo can even replace it
[20:12:36] <deep42thought> maybe we could even introduce a transition-package for the mirrorlist ... hmmm
[20:12:59] <tyzoid> deep42thought: The transition package would require that they've already moved to our mirrors
[20:13:09] <tyzoid> unless arch mainline hosts the transition package
[20:13:17] <tyzoid> for mirrors, that is
[20:13:18] <deep42thought> no, someone could put it into community, or are there tight restrictions for that?
[20:13:28] <deep42thought> right
[20:13:50] <brtln> you can just host it on any HTTP server
[20:13:56] <eschwartz> You could have people pacman -U http://mirror.archlinux32.org
[20:14:04] <brtln> and tell people to -U http://fobar for migration
[20:14:16] <deep42thought> they need manual intervention anyway, I guess
[20:14:24] <eschwartz> pacman -U will check the signatures and everything
[20:14:33] <brtln> exactly
[20:14:40] <deep42thought> well, if Polichronucci or City-busz sign the package, then it's fine
[20:14:59] <deep42thought> yeah, I referred to setting up a new mirrorlist, etc.
[20:15:07] <brtln> unless I missed something only City-busz is TU
[20:15:22] <deep42thought> also our package versions will differ and pacman will throw a lot of "local is newer" warnings
[20:15:36] <deep42thought> brtln: you might be right
[20:15:39] <brtln> that's fine, -Syuu will manage
[20:15:49] <eschwartz> easy enough; pkgrel++
[20:16:01] <brtln> and either way, I'm fine signing your transition packages too
[20:17:52] <deep42thought> eschwartz: pkgrel++ won't help if our version is smaller than mainline (because the compilation has not finished yet)
[20:20:50] <deep42thought> brtln: I still don't see, why three TUs signing our three master keys would not let pacman trust our packages - are there some resources where I can read about the trust model pacman uses?
[20:23:37] <eschwartz> deep42thought: pkgver in the arch packages is based on release date
[20:24:15] <eschwartz> and gpg trust allows several trusted keys to marginally trust a key for a combo full trust. But that doesn't let them sign more keys in return
[20:25:05] <deep42thought> so the combo full trusted key is not marginally trusted to sign?
[20:25:06] <eschwartz> we'll need our own pacman keyring
[20:25:21] <deep42thought> we have, I just wanted to understand the internals
[20:29:26] <tyzoid> deep42thought: FYI, City-busz has yet to sign my key
[20:29:34] <deep42thought> I know
[20:29:39] <deep42thought> I check it regularly
[20:29:57] <deep42thought> because when he signs, I can update archlinux32-keyring and you can start crunching ;-)
[20:33:35] <deep42thought> btw: It seems I was confused by owner-trust vs. trust in a key
[20:58:56] -!- eschwartz has quit [Remote host closed the connection]
[20:59:57] -!- eschwartz has joined #archlinux-ports
[21:01:13] <tyzoid> deep42thought: City-busz has signed the key :)
[21:01:51] <deep42thought> :-)
[21:03:50] <eschwartz> yay, I'm back and not segfaulted.
[21:04:20] <eschwartz> brtln: you totally can't hear me moaning about gcc 7, right?
[21:07:00] <brtln> what? I can't hear you
[21:07:26] <brtln> did someone say anything?
[21:07:50] * tyzoid pictures brtln standing by a large waterfall yelling
[21:08:18] -!- dmakeyev has joined #archlinux-ports
[21:10:08] <brtln> eschwartz: anyway, don't expect me to do anything magic about your issue
[21:10:11] <brtln> bring it to flyspray if you have to
[21:17:58] fsckd is now known as Father
[21:19:39] -!- Father has quit [Quit: bbl]
[21:38:37] <eschwartz> brtln: I was thinking of filing a mozilla bugreport, and maybe an "upstream issues" on flyspray. But I really have no idea whose fault it would be, except that gcc works when not applied to mozilla, so... ¯\_(ツ)_/¯
[21:45:46] <brtln> heftig maintains it and knows more gcc internals than I do so you can't lose anything
[21:45:53] <brtln> "it" == thunderbird
[21:48:56] -!- fsckd has joined #archlinux-ports
[22:04:09] -!- fsckd has quit [Quit: SIGTERM, gotta run. Bye!]
[22:04:39] -!- fsckd has joined #archlinux-ports
[22:10:43] <eschwartz> https://bugzilla.mozilla.org
[22:10:46] <phrik> Title: 1376908 - [gcc 7] crash when trying to connect to IRC (at bugzilla.mozilla.org)
[22:11:47] <tyzoid> eschwartz: gcc7 has support for IRC?
[22:12:10] <eschwartz> tyzoid: it is a bug on the thunderbird buzilla...
[22:12:21] <tyzoid> eschwartz: I'm looking at the bug
[22:12:33] -!- r00t^2_ has joined #archlinux-ports
[22:12:41] <tyzoid> I'm not sure what you mean by 'when switching to the Chat tab and connecting to IRC'
[22:12:43] <tyzoid> via what software?
[22:12:56] <tyzoid> web r00t^2
[22:13:20] <r00t^2_> hrm?
[22:14:01] <tyzoid> whoops, s/web/wb/
[22:14:05] <deep42thought> tyzoid: I think he has irc in thunderbird, which crashes if tb is compiled with gcc7
[22:14:25] <tyzoid> ah, so the build is successful
[22:14:48] -!- r00t^2 has quit [Ping timeout: 240 seconds]
[22:14:49] <tyzoid> eschwartz: You may want to reword that, I read it as the build was failing
[22:15:08] -!- isacdaavid has joined #archlinux-ports
[22:18:14] <r00t^2_> tyzoid: heh thanks
[22:18:56] r00t^2_ is now known as r00t^2
[22:21:55] <eschwartz> tyzoid: I suppose the bugzilla folks will know what I mean since it is filed against the Thunderbird project and the Instant Messaging component. As for the build failing, I did write build+run, but anyway it is too late to reword it -- bugzilla doesn't allow editing a comment.
[22:22:12] <tyzoid> ah
[22:22:37] <tyzoid> eschwartz: Fair enough. I didn't even realize thunderbird had an irc client, despite running it for the past four years
[22:23:00] <eschwartz> you can /ctcp version me if you don't believe it. :D
[22:23:11] <eschwartz> (Though I could be lying)
[22:23:24] <tyzoid> It says Thunderbir
[22:23:45] <tyzoid> Thunderbi*
[22:23:50] <tyzoid> I had seen that
[22:23:55] <tyzoid> but I didn't 2+2
[22:24:08] <tyzoid> * oblivious user is oblivious *
[22:24:28] <eschwartz> !grab tyzoid
[22:24:28] <phrik> eschwartz: Tada!
[22:27:28] -!- deep42thought has quit [Ping timeout: 240 seconds]
[22:30:44] -!- deep42thought has joined #archlinux-ports
[22:30:48] <tyzoid> I'll be back in an hour or so
[22:31:09] <tyzoid> deep42thought: Not sure if you saw the logs earlier, but I'm going to take a look at openqa
[22:31:18] <deep42thought> yeah, read that
[22:31:46] <tyzoid> also deep42thought, when are we planning on getting an iso build out?
[22:32:02] <tyzoid> and are we going to use the archlinux32 mirrors this time around?
[22:34:29] -!- r00t^2 has quit [Quit: i don’t know why i think pressing ctrl-c harder will help.]
[22:35:47] <deep42thought> I built one and put it into testing
[22:35:50] -!- r00t^2 has joined #archlinux-ports
[22:35:55] <deep42thought> I wouldn't yet trust our mirrors :-/
[22:36:15] <deep42thought> there are some broken package builds due to broken dependent packages
[22:36:32] <tyzoid> I built an iso out of our mirrors earlier and it installs just fine
[22:36:46] <deep42thought> hmm, then just switch
[22:36:56] <tyzoid> Sounds good
[22:37:26] <deep42thought> however, I would packages consider "unstable", testing packages "highly unstable" and staging packages "beyond possible repair" ;-)
[22:37:27] <tyzoid> anyway, see you in a little bit
[22:37:30] <deep42thought> cu
[22:37:45] -!- tyzoid has quit [Quit: WeeChat 1.8]
[22:49:08] -!- dmakeyev has quit [Ping timeout: 240 seconds]