#archlinux-ports | Logs for 2017-06-14

Back
[00:12:02] -!- H-H-H has quit [Ping timeout: 260 seconds]
[00:26:33] -!- H-H-H has joined #archlinux-ports
[00:32:45] -!- Faalagorn has quit [Remote host closed the connection]
[01:42:32] -!- yokel has quit [Ping timeout: 245 seconds]
[01:45:10] -!- yokel has joined #archlinux-ports
[02:02:39] -!- H-H-H has quit [Ping timeout: 240 seconds]
[02:18:05] eschwartz is now known as rfpujnegm
[02:20:56] rfpujnegm is now known as ztrawhcse
[04:12:28] guys is now known as eschwartz
[04:18:56] eschwartz is now known as guys
[05:25:19] ztrawhcse is now known as eschwartz
[05:37:48] eschwartz is now known as ztrawhcse
[05:38:15] ztrawhcse is now known as eschwartz
[06:21:33] -!- deep42thought has joined #archlinux-ports
[06:43:27] <deep42thought> ok, the build process is finished as far as it can: I have 580 broken packages left, as well as 2437 tasks waiting for their (broken) dependencies
[06:44:00] <deep42thought> if anyone wants to have a look into the broken packages: they're at https://eckner.net
[06:44:54] <deep42thought> the first hash is the one of the respective repository (packages/community), the second is the one of packages32 (irelevant for these packages)
[06:48:17] <tyzoid> This is for the arch32 mirror?
[06:49:03] <deep42thought> sry, I don't understand the question
[06:49:11] <deep42thought> these are the packages built for the mirror
[06:49:24] <deep42thought> for the 32-bit mirror
[06:50:56] <tyzoid> Okay. Wasn't sure whether you were referring to the packages that would go on the archiso or on the mirror
[06:51:56] <deep42thought> and btw: "gens" is blacklisted (automatically), because it falsely claims to depend on lib32-gtk, lib32-mesa-libgl and lib32-sdl
[07:27:14] <deep42thought> I uploaded a more detailed build master status at eckner.net/build-master-status
[07:40:52] -!- deep42thought has quit [Remote host closed the connection]
[08:36:19] -!- deep42thought has joined #archlinux-ports
[09:06:40] <tyzoid> deep42thought: In theory, if these build jobs were happening in parallel, you could throw it on a rabbitMQ store and dish builds out to multiple servers
[09:07:21] <deep42thought> you can already hand them out to multiple servers
[09:07:27] <deep42thought> (I'm using 2 at the moment)
[09:08:00] <tyzoid> how does it do the partitioning of the jobs?
[09:08:05] <tyzoid> alphabetical?
[09:08:11] <deep42thought> by dependencies
[09:08:27] <deep42thought> no job is assigned whose dependencies are still to be built
[09:08:40] <tyzoid> so what does the assigning?
[09:08:48] <tyzoid> just curious
[09:09:09] <deep42thought> the assigning is done by a bash script
[09:09:20] <tyzoid> Okay
[09:09:21] <tyzoid> Also, should we bring someone over from the #archlinux channel to see if they can't help us with the segfault?
[09:09:37] <tyzoid> there may be people over there that have an idea of what's going on.
[09:09:50] <deep42thought> yeah
[09:10:04] <deep42thought> to me, it sounds a lot like a problem between the kernel and the vm
[09:10:12] <tyzoid> Alright. I've got a quick article to read and a 1p paper to write
[09:10:18] <tyzoid> so I'll be done in about 20 mins
[09:10:21] <deep42thought> k
[09:37:25] <tyzoid> deep42thought: back
[09:38:18] <deep42thought> regarding your question about the build procedure: it's basically all in github.com/archlinux32/builder
[09:38:37] <deep42thought> these are some bash scripts to organize the build master, build slaves and the package database
[09:39:25] <tyzoid> Sweet
[09:39:33] <tyzoid> I'll need to give that a look over when I get some time
[09:39:56] <tyzoid> want to hop over to #archlinux?
[09:40:04] <deep42thought> feel free to report back any bugs/comments to me
[09:40:08] <deep42thought> yeah, ok
[10:39:10] -!- eroken_ has joined #archlinux-ports
[10:42:20] <tyzoid> deep42thought: If you're still online, I've got the vagrant file online
[10:42:21] <tyzoid> http://dl.tyzoid.com
[10:42:43] <deep42thought> I am
[10:42:51] <tyzoid> If you install the vagrant package
[10:43:12] <tyzoid> you can download the vagrantfile to an empty directory, and type 'vagrant up'
[10:43:18] <tyzoid> it'll automatically download/configure the box for you
[10:43:33] <deep42thought> sounds nice, thanks for the intro
[10:43:41] <tyzoid> and if you ever need to nuke it to get a clean box again, 'vagrant destroy && vagrant up' does that
[10:43:44] <tyzoid> vagrant halt shuts down
[10:43:54] <tyzoid> there's more, but man vagrant + google is helpful
[10:44:02] <deep42thought> yes, as usual ;-)
[10:44:47] <tyzoid> anyway, I'm off for the night. Let me know if you get anywhere, and we can resume this later.
[10:45:06] <deep42thought> thanks for all the help!
[10:45:23] <tyzoid> no problem. Can't be a team project if there's only one person :)
[10:45:28] <deep42thought> :-D
[10:45:37] -!- tyzoid has quit [Quit: Leaving]
[11:07:14] -!- eroken_ has quit [Ping timeout: 268 seconds]
[11:11:15] -!- AndrevS has joined #archlinux-ports
[12:13:38] -!- AndrevS has quit [Quit: umount /dev/irc]
[14:19:24] -!- Faalagorn has joined #archlinux-ports
[14:26:00] -!- Faalagorn1 has joined #archlinux-ports
[14:27:12] -!- Faalagorn has quit [Ping timeout: 260 seconds]
[16:11:47] -!- deep42thought has quit [Quit: Leaving.]
[18:18:26] -!- deep42thought has joined #archlinux-ports
[18:30:22] -!- tyzoid has joined #archlinux-ports
[18:31:39] <tyzoid> Hey deep42thought, anything happen while I was away?
[18:36:33] <deep42thought> tyzoid: not with the iso
[18:37:11] <deep42thought> I was busy removing some bugs from the build scripts and doing some actual work ;-)
[18:38:10] <deep42thought> my idea for the iso would be to use the lts kernel - that should have the least issues regarding its dependencies
[18:38:18] <tyzoid> Possibly
[18:38:28] <tyzoid> But I was thinking we could use our own 'linux' package as well
[18:38:34] <tyzoid> because we have our 32bit port of it.
[18:39:03] <deep42thought> so you'd just turn back the version number of our own package?
[18:39:47] <tyzoid> Not sure why that would be necessary. Don't we have a linux-*-i686.tar.xz package?
[18:39:56] <tyzoid> one that is built as part of your package build system?
[18:40:07] <deep42thought> yes
[18:40:17] <deep42thought> but how is that better/different from the original one?
[18:40:35] <tyzoid> Well I thought the idea was to determine if there was an issue with the original linux package
[18:40:56] <tyzoid> using one that has gone through our own build system, that we have verified to work by building a system with it, may work
[18:40:59] <tyzoid> or at least aid in debugging
[18:41:10] <deep42thought> actually, we have 4.10.13 in core and 4.11.3 in staging
[18:41:50] <deep42thought> so we could even use different versions :-D
[18:42:01] <tyzoid> All valid possibilities
[18:42:21] <tyzoid> At this point, my suspicion would be a broken kernel package somehow
[18:42:30] <tyzoid> so that's why I'm thinking this would be a good place to start testing
[18:42:43] <tyzoid> Did the vagrant setup work for you, btw?
[18:43:12] <deep42thought> haven't tested it yet, sry
[18:43:18] <tyzoid> No problem
[18:43:31] <tyzoid> Once we get the iso building again, I'll write some documentation for the vagrant setup
[18:43:41] <deep42thought> but it sounds straight forward enough that I should not fail ;)
[18:44:11] <tyzoid> I have no doubts about that :)
[18:45:01] <tyzoid> The other thing that we can look at, once we get farther along in the processs, is that vagrant can run setup/provisioning scripts
[18:45:14] <tyzoid> and we could configure it to build the dual or i686 packages there automatically
[18:45:31] <tyzoid> Anyway, I'm heading out to lunch.
[18:45:38] <tyzoid> I'll be back in about 30-45min.
[19:12:17] <tyzoid> Back
[19:14:51] <deep42thought> I'm still at dinner
[19:14:58] <deep42thought> should be done in ~10 mins
[19:15:13] <tyzoid> np
[19:15:18] <tyzoid> Take your time
[19:15:26] <tyzoid> I eat quickly as a bad habit :)
[19:17:30] <deep42thought> me, too, but at breakfast and dinner, I'm slowed down by two children ...
[19:24:35] <deep42thought> so, I'm done
[19:40:25] <tyzoid> Let me know if you have time to look at the custom package thing
[19:40:33] <tyzoid> I'm grabbing the vm on my current machine
[19:41:14] <deep42thought> I'm waiting for one build process to finish, so I can reboot (upgraded the kernel on the server) and then I'll try to generate all kinds of isos :-D
[19:41:39] <tyzoid> nice
[19:44:07] <deep42thought> the compilation of qt5-webkit takes like forever ^^
[19:45:38] <tyzoid> Yup
[19:50:46] <deep42thought> the 4.10.13 kernel is from before 28th april, so it should not have any bug introduced in the kernel for the june image
[19:51:48] <tyzoid> is that the one we have in our repo?
[19:52:12] <deep42thought> this is the one I copied from the official mirrors into our database, so we'd had something to start at
[19:52:14] <tyzoid> If so, couldn't we just start packaging the isos with our 32bit repo instead?
[19:52:19] <tyzoid> ah
[19:52:45] <deep42thought> currently all packages in our stable repositories except archlinux32-keyring and pacman are "original" packages
[19:53:01] <tyzoid> ah
[19:53:04] <tyzoid> makes sense
[19:53:10] <deep42thought> but they're old
[19:53:19] <deep42thought> as you can see from the kernel
[19:53:25] <tyzoid> so the question then becomes when we want to start moving the iso builds to our repo then
[19:53:37] <deep42thought> our packages just went up to "testing" (and "community-testing")
[19:53:47] <deep42thought> yes, would make sense
[19:54:05] <deep42thought> at least to see if the old kernel still boots
[19:54:15] <deep42thought> and if it does, there is actually no reason to move back ...
[19:54:33] <tyzoid> I'll test that then
[19:55:00] <deep42thought> so you'll build the iso yourself?
[19:55:20] <deep42thought> that'll accellerate it, because I don't need to wait for the build+reboot
[19:55:33] <tyzoid> Yeah, I just downloaded the vm to do the builds
[19:55:41] <deep42thought> ok, nice
[19:55:54] <tyzoid> btw, the vm requires the virtualbox extension pack
[19:56:01] <tyzoid> I forgot to disable usb2, so...
[19:56:12] <deep42thought> :-D
[19:57:32] <tyzoid> Okay, build started
[19:57:40] <tyzoid> let's hope it can grab the packages alright
[20:11:47] <deep42thought> any complaints about missing packages and/or signatures so far?
[20:12:50] <tyzoid> Nope
[20:12:55] <tyzoid> just tested the iso, boots fine
[20:13:02] <deep42thought> good
[20:14:21] <tyzoid> just needed to modify the /usr/share/archiso/configs/releng/pacman.conf file to use the archlinux32 mirrors instead of /etc/pacman.d/mirrorlist
[20:14:37] <tyzoid> that approach won't necessarily work for the dual iso
[20:14:46] <tyzoid> but I'm only generating the i686 iso for now
[20:15:01] <deep42thought> you can just insert the 32-bit only mirror before your dual-arch mirror
[20:15:17] <deep42thought> in your regular mirrorlist
[20:15:21] <tyzoid> Though I imagine that we won't be supporting the dual isos once mainline finally drops support
[20:15:29] <tyzoid> deep42thought - That might work
[20:15:34] <tyzoid> if it does, sweet
[20:15:35] <deep42thought> I personally like the dual boot iso
[20:15:55] <deep42thought> one usb thumb drive to rule them all
[20:16:04] <tyzoid> Do you think there's a benefit of us maintaining that once the official support is finally dropped?
[20:16:19] <deep42thought> I have personal benefit
[20:16:21] <deep42thought> ;-)
[20:16:33] <deep42thought> and it's not been hard, so far
[20:17:05] <deep42thought> but if it'll get hard, we can drop it at any time
[20:19:57] <deep42thought> maybe someone else also likes to have one usb thumb drive to rescue/install both i686 and x86_64 arch
[20:25:44] <tyzoid> Fair enough
[20:26:14] <tyzoid> So you're working on getting the builds started for updating the repo?
[20:26:22] <deep42thought> yep
[20:27:55] <deep42thought> https://eckner.net
[20:28:58] <deep42thought> I just gave all "broken" packages a second chance - and some of them do not fail to build anymore ...
[20:29:12] <deep42thought> maybe the sources reappeared
[20:30:28] <tyzoid> hm
[20:30:31] <tyzoid> that's odd
[20:31:10] <deep42thought> a lot of failed builds are due to missing sources
[20:31:21] <tyzoid> like git repos no longer there?
[20:31:37] <deep42thought> 404's, 40x's, ...
[20:31:47] <tyzoid> ...
[20:32:09] <tyzoid> so the question is how arch builds them then
[20:32:09] <deep42thought> I even learned about some new (for me) http error codes ...
[20:32:13] <tyzoid> lol
[20:32:14] <tyzoid> http418?
[20:32:19] <deep42thought> it did in the past, I guess
[20:32:39] <deep42thought> that one I knew already ;-)
[20:32:47] <tyzoid> 402?
[20:33:00] <tyzoid> 402 is a fun one
[20:33:14] <tyzoid> still compliant with the GPL :P
[20:34:03] <deep42thought> yep, that one
[20:35:46] <deep42thought> I should have read the full description of the code instead of relying on its message ...
[20:37:40] <deep42thought> so we need more build slaves - just to spread the downloads from the google code api over more ip addresses ...
[20:38:18] <tyzoid> depending on space requirements, I could volunteer my mirror to help do builds
[20:38:36] <tyzoid> also dependson whether it requires arch to do the builds
[20:38:40] <tyzoid> my mirror runs on ubuntu
[20:38:55] <deep42thought> actually, I'd like to have a little more stable build scripts before distributing among "foreign" machines
[20:39:06] <deep42thought> and yes, it requires arch
[20:39:08] <tyzoid> sounds good
[20:39:11] <tyzoid> drats
[20:39:44] <tyzoid> Doesn't arch have a sources library on the mirror?
[20:39:45] <deep42thought> but maybe a systemd-container with arch inside could be enough - not sure, though
[20:40:03] <tyzoid> neither am I. That said, i've got ubuntu with upstart
[20:40:07] <deep42thought> yes, but only for packages where the license requires them to publish the source
[20:40:08] <tyzoid> so that wouldn't work regardless
[20:40:23] <tyzoid> Could be at least some help
[20:40:36] <deep42thought> waht could be some help?
[20:40:42] <tyzoid> if the sources aren't available
[20:40:48] <tyzoid> we could always cross-check there
[20:40:51] <deep42thought> you think like I: sources.archlinux.org is already a regular fallback
[20:40:56] <tyzoid> ah
[20:40:59] <tyzoid> well then
[20:41:10] <tyzoid> great minds think alike
[20:41:36] <deep42thought> aren't great minds also humble?
[20:41:41] <deep42thought> ;-P
[20:41:59] <tyzoid> Some are
[20:43:01] <brtln> tyzoid: we just don't build these
[20:43:29] <brtln> so nobody even knows that sources are not there
[20:43:51] <brtln> we lack a CI unfortunately
[20:47:31] <deep42thought> I think, there is no need to recompile all old packages - I see it merely as a test for the build system
[20:48:01] <deep42thought> if one of the packages receives an upgrade, that one should be buildable - e.g. no missing sources, ...
[20:48:16] <tyzoid> deep42thought: sounds like a decent approach
[20:48:49] <tyzoid> brtln: I didn't know ABS had a CI system
[20:49:26] <brtln> no, we don't have it
[20:49:37] <brtln> that's why so many packages fail
[20:49:40] <tyzoid> :(
[20:50:33] <deep42thought> is there actually any verification that a PKGBUILD works at the moment it is upgraded?
[20:50:56] <deep42thought> e.g. besides by the maintainer who compiled a package from it?
[20:51:41] <tyzoid> Not sure if any of them have automated tests
[20:52:00] <deep42thought> so it's responsibility of the maintainer
[20:52:04] <tyzoid> I assume if the build works, the package should in theory work
[20:52:12] <tyzoid> but that's between dev and maintainer, like you said
[20:52:22] <tyzoid> Arch usually has a lag time on the order of days with that, though
[21:14:13] -!- titus_livius has joined #archlinux-ports
[21:14:18] <tyzoid> But I'm wondering if there's anything that could help us debug this further
[21:14:34] <tyzoid> drathir: Do you have a 32bit live cd?
[21:14:39] <drathir> yep fresh updated from working ones i guess cood try check from what last version working....
[21:15:11] <tyzoid> drathir: Which version of the livecd are you using?
[21:15:25] <drathir> yea have but ekhm old ;/
[21:15:35] <tyzoid> The lastest working one is http://32.arlm.tyzoid.com
[21:15:48] <tyzoid> 2017.06.* fails on boot with a segfault
[21:15:51] <tyzoid> I assume for a similar reason
[21:16:00] <tyzoid> but it shouldn't matter
[21:16:02] <drathir> 4.8.6-1 ;p
[21:16:12] <tyzoid> is that your kernel version?
[21:16:28] <drathir> k i will extract from that one, and thats live cd used...
[21:16:49] * drathir only use it for booting anyway...
[21:17:07] <drathir> like sayed ekhm old one ^^
[21:17:12] <tyzoid> drathir: We've had good success with http://32.arlm.tyzoid.com
[21:17:37] <tyzoid> You know how to install it?
[21:17:38] <drathir> k downloading
[21:17:44] <deep42thought> I have most of the intermediate kernels in my pacman cache
[21:17:54] <tyzoid> deep42thought: cool
[21:18:09] <tyzoid> Unless you have any other thoughts, I'm just going to have him try pacman -U with this one
[21:18:29] <deep42thought> yes
[21:18:39] <deep42thought> if it works, bisecting to the breaking change would be nice
[21:18:57] <tyzoid> well I'd rather not have him do too many reinstalls, since this is an actual installation
[21:19:08] <tyzoid> we could bisect on our iso building easier than doing it to a real system
[21:19:13] <deep42thought> drathir: you might even have a lot of intermediate kernels in your pacman cache
[21:19:40] <deep42thought> tyzoid: good point
[21:20:13] <drathir> k good idea with cache hopefully not cleaned and last there is 4.10.11-1 but not sure if cleaned cache it could be latest one ?
[21:20:45] <tyzoid> if it's not cleaned, the lastest one would be the one installed
[21:20:50] <tyzoid> so you'd want to grab the one before that
[21:21:01] <deep42thought> drathir: 4.10.11 should work for you - at least you were able to install a new kernel afterwards ;-)
[21:21:13] <tyzoid> but our testing shows up to 4.10.13 should work
[21:21:28] <drathir> not sure if cleaned after last upgrades its a on the go external usb install
[21:21:36] <drathir> not often used...
[21:22:03] <drathir> yea lets check need reboot...
[21:22:24] <tyzoid> Let us know how that goes
[21:28:04] <tyzoid> hey ddep42thought: The discussion in the #archlinux channel brought up https://bugs.archlinux.org
[21:28:11] <tyzoid> deep42thought*
[21:28:22] <tyzoid> looks to be a possible cause of the segfault issue
[21:28:33] <tyzoid> so it might not be a problem with the linux package, but with glibc
[21:38:01] -!- fhdrin has joined #archlinux-ports
[21:43:21] <tyzoid> deep42thought: just making sure you saw my message about the glibc thing
[21:43:29] <deep42thought> sry, was briefly afk
[21:43:32] <tyzoid> np
[21:46:42] <deep42thought> honestly, I have no clue, what the issue is
[21:47:26] <deep42thought> so it might be glibc as well
[21:51:20] <tyzoid> deep42thought: Could you hop in #archlinux for a min?
[21:51:52] <deep42thought> sure
[22:00:37] -!- eschwartz has quit [Remote host closed the connection]
[22:01:06] -!- eschwartz has joined #archlinux-ports
[22:02:33] <tyzoid> deep42thought: I saw some of that discussion happening and thought it was important
[22:02:58] <tyzoid> so basically that's the arch devs saying that they're dropping support early without telling anyone :/
[22:03:07] <deep42thought> yep
[22:03:25] <deep42thought> but honestly, we don't care, because the build system is _almost_ up and running
[22:04:06] <tyzoid> Okay. So should we shoot for mid august to be fully operational with our systems, then?
[22:04:26] <deep42thought> I'd try to get it running as soon as possible
[22:04:37] <tyzoid> We should probably hammer out some sort of schedule to publish on our website
[22:05:18] <deep42thought> I don't think that's a good idea
[22:05:38] -!- dmakeyev_ has joined #archlinux-ports
[22:05:40] <deep42thought> if we have something running, we can spread the news
[22:05:54] <deep42thought> but if we set up a schedule we can't meet, that's bad
[22:06:04] <drathir> k than can confirm that testing update fixing a issue...
[22:06:44] <tyzoid> deep42thought fair enough
[22:06:48] <drathir> on Linux arch_i386_mobi 4.11.4-1-ARCH #1 SMP PREEMPT Fri Jun 9 08:07:52 CEST 2017 i686 GNU/Linux
[22:07:44] <tyzoid> drathir: I missed part of the converstation about the glibc stuff
[22:07:47] <drathir> but needed chroot and upgrade from live iso sadly if error occured...
[22:07:48] <tyzoid> what was the ultimate fix?
[22:08:08] <tyzoid> drathir: not sure I understand
[22:11:21] <drathir> guys im wonder after oficially arch stop updatind i686 if the compatibility with #archstrike will be keeped?
[22:11:33] <tyzoid> I'm not familiar with archstrike
[22:11:39] <drathir> tyzoid: need to u[grade to testing/mkinitcpio-busybox 1.25.1-2 [installed]
[22:12:13] <tyzoid> drathir: Given that archstrike supports arm, I wouldn't think they'd drop support for i686
[22:13:40] <drathir> tyzoid: but #archstrike is onry repository level of packages which was on top of archlinux...
[22:14:02] <tyzoid> I don't see why it won't be able to be moved on top of archlinux32's repos
[22:14:08] <deep42thought> drathir: we try to kkep our packages as close as possible to the x86_64 packages
[22:14:23] <tyzoid> though it couldn't hurt reaching out to them
[22:14:24] <drathir> k thats good to hear... better always to ask...
[22:14:34] <tyzoid> thanks for bringing it up
[22:14:38] <tyzoid> Good to keep in mind
[22:16:07] <drathir> deep42thought: i wish of that, bc for sure that ppl will like, not like some other arch linux based distros trying doing everything by own way even including own package managers ;/
[22:17:05] <deep42thought> I'm not happy with all the decisions, arch made so far, but it'll be the least effort trying to stick close to it
[22:18:11] <drathir> deep42thought: sadly in my opinion all started move to worse with decision of follow systemd... that was first and the biggest archlinux mistake in my opinion...
[22:18:28] <drathir> byt that only mine personal opinion...
[22:18:36] <deep42thought> (mine, too)
[22:19:27] <tyzoid> I just hope busybox continues to support i686 for the time being
[22:19:32] <tyzoid> systemd*
[22:19:46] <tyzoid> otherwise we'd be royally screwed
[22:20:21] <deep42thought> the builds work without virtualization
[22:20:34] <deep42thought> only the testing would get nasty
[22:20:44] <tyzoid> deep42thought: package builds? or iso builds?
[22:20:48] <deep42thought> both
[22:21:01] <deep42thought> it's solely systemd containers
[22:21:02] <drathir> im not hidding that dont like systemd whole idea and think its biggest security hole ever created in linux, but from sentiment to archlinux im still trying to stick to it, but hpnestly if things go really really badly in my opinion only one alternative this days similar to archlinux from old days is bsd...
[22:21:03] <deep42thought> and setarch
[22:21:06] <tyzoid> I'd still recommend virtualization for them, just to have a clean environment
[22:21:29] <tyzoid> but the containers should work properly regardless
[22:21:49] <deep42thought> I'm currently building in containers
[22:22:43] <tyzoid> I'm partial towards the vms, but it might just be that the containers are quicker to create/destroy
[22:25:38] <tyzoid> So deep42thought: after looking at it further, it looks like the glibc bug only affected busybox
[22:25:54] <deep42thought> hmpf
[22:30:58] <deep42thought> have you tried if the iso boots if built from our testing packages?
[22:31:02] <deep42thought> (e.g. new kernel)
[22:31:09] <tyzoid> I didn't try the testing repos
[22:31:29] <deep42thought> we should somehow identify the responsible package
[22:36:43] <tyzoid> deep42thought: Starting the build vm up again
[22:37:38] <drathir> may i askwhich repo names You are planning to use?
[22:37:57] <drathir> archlinux32-core?
[22:38:01] <deep42thought> core community extra
[22:38:02] <deep42thought> no
[22:38:08] <deep42thought> the same as the original ones
[22:38:16] <deep42thought> the distinction will be the architecture
[22:38:44] <drathir> hmmm... You think there will be no ppl confused?
[22:39:11] <drathir> eg in troubleshooting which one archlinux or archlinux32 repos are used?
[22:39:16] <deep42thought> that's the easiest way: you switch your mirror, update your keyring and you're settled with the new package database
[22:39:40] <deep42thought> from november on there is no support for i686 in archlinux
[22:40:01] <deep42thought> that would be like reporting bugs for archlinux on arm to archlinux instead of archlinuxarm
[22:42:11] <drathir> in theory as add it as separaate naming it would get it over the even untouched archlinux repositories bc of version bump... that in theory in my opinion have sense if archlinux will allow to sync all 32bit mirrors to archlinux32...
[22:42:31] <drathir> lets see how the migration would be looks like...
[22:43:21] <deep42thought> I don't think, our packages will land on official archlinux mirrors alongside with x86_64 package
[22:43:22] <deep42thought> s
[22:43:36] <deep42thought> much like with packages for arm
[22:43:47] <tyzoid> Building with the testing repo enabled
[22:43:53] <deep42thought> :-)
[22:44:35] <tyzoid> I agree with deep42thought here
[22:44:45] <tyzoid> The whole idea is to make this organizationally like archlinux-arm
[22:45:47] <tyzoid> we'd be independent, but with the endorsement of archlinux
[22:45:54] <tyzoid> As was mentioned in #archlinux:
[22:45:59] <tyzoid> grazzolini | tyzoid: I'm not saying it won't be at minimum endorsed in the future. we've created #archlinux-ports
[22:46:07] <drathir> in that case i think would be nice separate naming from archlinux to avoid migration issues like someone still using archlinux repositories, eg providing archlinux32-keyring archlinux32-mirrorlist eg like that...
[22:46:23] <tyzoid> we do have archlinux32-keyring :)
[22:46:42] <tyzoid> and I agree on mirrorlist too
[22:46:49] <tyzoid> but there's not too many packages like taht
[22:48:30] <deep42thought> why do you want to keep both mirrorlists?
[22:48:32] <tyzoid> Alright, so deep42thought: for some reason, I needed to run it three times, but it's finally building an iso
[22:48:43] <deep42thought> :-/
[22:48:56] <tyzoid> deep42thought: During the transition time, I think it would be nice.
[22:49:01] <tyzoid> up to you on that one, though
[22:49:15] <tyzoid> just make archlinux32-mirrorlist conflict with archlinux-mirrorlist
[22:49:53] <tyzoid> That said, the mirrorlists are usually installed to mirrorlist.pacnew for most people anyway, as they modify theirs
[22:50:13] <deep42thought> hmm, haven't thought of that - only of a separate file "/etc/pacman.d/mirrorlist32" as well
[22:50:39] <tyzoid> My thought is that it would allow people to switch more easily back to the mainline if they have issues
[22:50:45] <tyzoid> your call, really.
[22:50:50] <tyzoid> I don't have much of an opinion there
[22:51:03] <drathir> and there for ppl migrating from archlinux could be only needed add archlinux32-core archlinux32-testing archlinux32-community archlinux32-community-testing archlinux32-extra wnd no colide, but i probably need to see migration wiki guide to visualize how it will looks like...
[22:51:29] <tyzoid> well drathir, we could make an AUR package to do the migration more easily if we really wanted to
[22:51:31] <tyzoid> but that's not necessary
[22:51:45] <tyzoid> you only need to modify your mirrorlist and download the keys to get it working
[22:51:55] <tyzoid> the AUR package would handle the keys for us
[22:52:39] <drathir> for me eg. no problem to edit pacman.conf to my needs im even keep mantain mine pacserve+repo for local usage for x86_64 one repos...
[22:53:17] <drathir> tyzoid: and for keys i assume packages and repo will be signed right?
[22:53:23] <tyzoid> yup
[22:53:28] <tyzoid> We wouldn't want to do anything unsigned
[22:53:34] <tyzoid> that's a security hole waiting to happen
[22:53:37] <drathir> thats great ++
[22:53:47] <deep42thought> well
[22:53:56] <deep42thought> I've been thinking about repo signing a lot
[22:54:00] <tyzoid> deep42thought: it's not that difficult to maintain
[22:54:05] <deep42thought> and honestly: I don't like how it's implemented
[22:54:08] <tyzoid> just have a master key sign dev keys like the arch project does
[22:54:19] <deep42thought> package signing is no problem
[22:54:22] <deep42thought> (is done already)
[22:54:34] <deep42thought> however _repository_ signing is something totally different
[22:54:50] <tyzoid> oh, I'm not aware of this.
[22:54:58] <tyzoid> Mind giving me a brief explanation?
[22:54:59] <deep42thought> I'm usually a friend of more encryption and more signing, but with repositories, I don't like it
[22:55:19] <deep42thought> well, you can also sign the core.db.tar.gz and alike
[22:55:28] * drathir wonder if its planned in future when all stabilize to support cjdns and dn42 ?
[22:55:31] <deep42thought> have a look at my repository at arch.eckner.net/archlinuxewe
[22:56:12] <deep42thought> the masterkeys.gpg contains two keys - one for the packages and one for the repository
[22:56:20] <drathir> signing is really matter even when downloading its easy if package sign not match there is corrupted or outdated...
[22:56:49] <deep42thought> drathir: as I said: regarding signing packages - yes
[22:57:31] <deep42thought> however if you really want to sign the package database and actually verify it, you need pacman to trust the database signing key in the same manner as the package signing keys
[22:57:50] <tyzoid> So... not sure what the benefit of repo signing is; if someone has the dev keys, what's to stop them from distributing those?
[22:58:22] <deep42thought> right
[22:58:25] <drathir> and all mirrors https supported... ?
[22:58:42] <deep42thought> one could set up a tainted mirror with a few outdated packages otherwise
[22:59:03] <tyzoid> drathir: https mirror doesn't really have any benefit
[22:59:04] <deep42thought> with signing, you'd need to stop syncing the repository completely
[22:59:08] -!- eschwartz has quit [Ping timeout: 240 seconds]
[22:59:13] <tyzoid> all of our packages are signed anyway
[22:59:27] <tyzoid> deep42thought: makes sense
[22:59:35] -!- guys has quit [Ping timeout: 246 seconds]
[23:00:43] <deep42thought> There's also a security risk if we would sign the database: We'd need to keep the private key somewhere online - e.g. on the build master or the master mirror
[23:01:27] <deep42thought> and if that key is stolen, you could put signed packages on a mirror and no one would notice, because pacman does not know a difference between package keys and database keys
[23:01:46] <tyzoid> deep42thought: the iso works from the testing repo
[23:01:54] <deep42thought> interesting
[23:02:11] <deep42thought> ah, wait
[23:02:15] <brtln> yes, we haven't solved that in a sane way so far so official databases are unsigned as well
[23:02:19] <deep42thought> the new kernel is in staging iirc
[23:03:28] <deep42thought> brtln: one way could be to keep the database signing keys in a separate pacman keyring, but I think it wouldn't be worth the effort
[23:04:01] <deep42thought> tyzoid: yep, it's in staging ...
[23:04:04] <brtln> Allan is contact person about it
[23:04:11] <brtln> he likes to torpedo every idea about it
[23:04:16] <deep42thought> :-D
[23:04:19] <tyzoid> ah, well.
[23:04:35] <brtln> I'm not even going to bring it up with him after last time :P
[23:05:05] <deep42thought> tyzoid: let me see, if it's unstageable already ...
[23:05:16] <tyzoid> Well I need to head out momentarily here
[23:05:20] <deep42thought> np
[23:05:21] <tyzoid> I can try to be back on in about 15 mins
[23:05:38] <deep42thought> don't hurry, I might go to bed anytime now ...
[23:06:23] <tyzoid> alright
[23:06:29] <tyzoid> well, I'm off
[23:06:33] <deep42thought> cu
[23:06:35] <tyzoid> If it catch ya, then I catch ya
[23:06:46] -!- tyzoid has quit [Quit: WeeChat 1.8]
[23:17:12] -!- H-H-H has joined #archlinux-ports
[23:17:50] <H-H-H> hmm seems this chanel isnt on my autojoin list on this system lol
[23:20:51] -!- tyzoid has joined #archlinux-ports
[23:21:02] <tyzoid> alright, back @deep42thought
[23:21:10] <deep42thought> I'm still here
[23:21:30] <deep42thought> and the unstaging is still in progress O.o
[23:22:46] <tyzoid> sounds good
[23:23:14] <deep42thought> well, usually it was a lot faster ...
[23:24:18] <deep42thought> ok, it's done, but still in staging
[23:24:34] <deep42thought> which means, that something not yet built depends on 'linux'
[23:25:19] <deep42thought> nvidia ...
[23:25:56] <tyzoid> god damn
[23:26:00] <tyzoid> nvidia :/
[23:27:51] <deep42thought> which is broken ...
[23:28:04] <tyzoid> Should we skip nvidia for now?
[23:28:38] <deep42thought> just move ahead and put staging next to testing (and community-staging next to community-testing) in the pacman.conf
[23:28:57] <deep42thought> this will potentially break everything, but the kernel should work
[23:33:12] <deep42thought> something is strange when building nvidia - pacman complains about missing dependencies, because it tries to download newer versions than available (on archlinux32.org)
[23:33:43] <deep42thought> ah, crap
[23:33:43] <tyzoid> Shouldn't the newest versions be available on the nvidia website?
[23:33:53] <deep42thought> archlinux packages
[23:33:58] <deep42thought> not source tar balls
[23:35:09] <deep42thought> I found my mistake: on this machine, I had only some dual-arch mirror in the mirrorlist
[23:35:54] <tyzoid> ah
[23:41:21] -!- xorond has joined #archlinux-ports
[23:42:12] <tyzoid> xorond: Also, this channel has irc logs available on the mirrors: https://mirror.archlinux32.org
[23:42:22] <xorond> sweet, thanks
[23:42:38] <tyzoid> let me or deepthought know if you need anything from us
[23:42:57] <xorond> will do
[23:52:27] -!- Polichronucci has quit [Ping timeout: 255 seconds]
[23:52:45] -!- Polichronucci has joined #archlinux-ports
[23:54:49] <deep42thought> ok, something seems to be wrong with the build system - and it can be seen with the nvidia package (it's mostly a note to myself). But I'll go to bed now.
[23:55:09] <tyzoid> alright. I'd say good work for today
[23:55:24] <tyzoid> quick question before you go
[23:55:40] <tyzoid> Is staging ready for doing some iso tests?
[23:55:45] <tyzoid> I can work on testing that tonight.
[23:56:45] <tyzoid> deep42thought, you still there?
[23:56:50] <deep42thought> yes
[23:57:10] <deep42thought> well, everything in staging was built successful, but it might lack some dependent packages
[23:57:27] <deep42thought> e.g. new python version might be in staging, but not all python-* packages
[23:57:37] <tyzoid> I don't really care much about python
[23:57:41] <deep42thought> I know
[23:57:42] <tyzoid> it's mostly the core stuff
[23:57:44] <tyzoid> like the linux package
[23:57:49] <deep42thought> just wanted to give an example
[23:57:50] <tyzoid> I'll test it, and let you know
[23:57:52] <tyzoid> thanks
[23:58:06] <deep42thought> I'll go now
[23:58:08] <deep42thought> cu later!
[23:58:11] <tyzoid> see ya
[23:58:13] -!- deep42thought has quit [Quit: Leaving.]
[23:58:38] -!- fhdrin has quit [Quit: Leaving]