Centurion_Dan | gnu_srs1: excellent. I'll pull and test later today. | 02:31 |
---|---|---|
Centurion_Dan | KatolaZ, golinux: Are we likely to get Devuan ASCII released before June 20 when Debian Jessie goes into LTS phase?? | 02:32 |
golinux | Centurion_Dan: I have no idea. NewGnuGuy and I were the only ones at yesterday's meet. | 02:40 |
golinux | Things have been really slow lately. | 02:40 |
Centurion_Dan | sorry I missed it. I have had a shit time the last couple of days... and wasn't able to attend. | 02:40 |
Centurion_Dan | I've got an abcessed 1st Molar... | 02:41 |
golinux | Ouch | 02:41 |
Centurion_Dan | Is their much in the way of RC bugs for ASCII?? | 02:46 |
Centurion_Dan | golinux: ? | 02:46 |
fsmithred | Centurion_Dan, I think we can do it by then | 02:47 |
fsmithred | the whole polkit thing is a little wonky, but there are ways to make everything work | 02:47 |
golinux | Centurion_Dan: I'm not the best one to ask as I don't really understand the *kit stuff very well | 02:48 |
Centurion_Dan | fsmithred: you reported odd consolekit situations the other day... is that something special desktop live or do you think it's an RC bug we still need to resolve?? | 02:55 |
fsmithred | It's in the rc | 03:13 |
fsmithred | not sure if there's a bug | 03:13 |
fsmithred | everything works. It's just an odd combination of things | 03:13 |
fsmithred | people do run into some troubles if they install the desktop in parts instead of using the task package | 03:14 |
fsmithred | and then once they get all the right packages, stuff works | 03:14 |
fsmithred | Centurion_Dan, correction - that was in a post-rc build from May 19. There's a newer version of policykit-1 in ascii-proposed that I should test. Can do it this weekend. | 04:58 |
fsmithred | sleep time | 04:59 |
Centurion_Dan | fsmithred: I often install without using the tasks. It's important that the packages themselves have correct dependencies. | 04:59 |
Centurion_Dan | We can tweak dependencies using overrides in amprolla.... | 05:00 |
fsmithred | oh really? | 05:00 |
fsmithred | so we can change the deps on the five or six desktop? | 05:00 |
Centurion_Dan | yup... not sure about amprolla3, although it should too... | 05:00 |
fsmithred | that sounds like a good idea | 05:01 |
fsmithred | I'm assuming it's easier than repackaging | 05:01 |
Centurion_Dan | would have to confirm with parazyd, but that was a key feature I helped nextime implement in original amprolla. | 05:02 |
fsmithred | ok, I'm gonna go have happy amprolla dreams | 05:02 |
fsmithred | g'night | 05:02 |
Centurion_Dan | of course binary dependencies still require package rebuilds... | 05:03 |
Centurion_Dan | sweet dreams ;-) | 05:03 |
KatolaZ | Centurion_Dan: ? | 07:50 |
KatolaZ | how is ascii related to jessie going LTS? | 07:51 |
gnu_srs1 | KatolaZ: (10:24:32 PM) srs: KatolaZ: Finally :) I've pushed an updated policykit-1 to ascii-proposed. Tested with xfce4, works fine ;) | 08:02 |
KatolaZ | ok gnu_srs1 | 08:34 |
KatolaZ | tnx | 08:34 |
Centurion_Dan | KatolaZ: Technically when debian jessie goes LTS that means support for Devuan Jessie is curtailed at least upstream... so either we step up when needed or we effectively have many probably unsupported packages in our current stable release. Ascii's release would reduce the tension as those who have bosses/clients being pressured to prove that they have se | 10:33 |
Centurion_Dan | security support would be able to say, there is a new Devuan release and we are planning an upgrade. | 10:34 |
Centurion_Dan | I actually have this very situation with a client at the moment, security support is important to them and if I can't assure them that the OS they're relying on for serving their solutions doesn't have security support, then that means there is a real risk their accountants will demand switching to an enterprise supported distro like ubuntu or redhat .... | 10:36 |
KatolaZ | Centurion_Dan: ASCII will be realsed when it's ready :) | 10:38 |
KatolaZ | and it seems to be almost ready | 10:39 |
Centurion_Dan | KatolaZ: Don't get me wrong, I know I haven't been able to help much this time around... and I think you people have done a very fine job indeed. So far it's been very good and a smooth upgrade where I've applied it, with minor caveats and those have been addressed. | 10:41 |
KatolaZ | Centurion_Dan: I am not getting you wrong or right | 10:41 |
KatolaZ | you asked a question, I gave you an answer | 10:41 |
KatolaZ | :) | 10:41 |
Centurion_Dan | :-) | 10:42 |
Centurion_Dan | Are there many RC bugs left holding a release back? | 10:42 |
KatolaZ | there are a few minor things that should be fixed by the latest upgrade of policykit | 10:43 |
KatolaZ | if that goes well, there should not be anything major left | 10:43 |
NewGnuGuy | Centurion_Dan: You are listed as the dbus maintainer. Is that correct? | 10:46 |
Centurion_Dan | I was chatting with fsimthred earlier about odd combinations of packages re consolekit and policykit which apparently occur if installing a system without using task meta packages. | 10:46 |
Centurion_Dan | NewGnuGuy: Yeah, I took it over at one stage... | 10:46 |
KatolaZ | Centurion_Dan: that's something that we will not fix | 10:47 |
KatolaZ | that's why task-*-desktop packages exist | 10:47 |
KatolaZ | we can't and shouldn't add dependencies on the polkit stuff | 10:47 |
KatolaZ | if people go "DIY", they know there will be bumps | 10:48 |
KatolaZ | there is documentation about which backend is recommended with each DE | 10:49 |
KatolaZ | and there are task-*-desktop meta-packages that provide everything needed | 10:50 |
Centurion_Dan | KatolaZ: :-( that's a poor approach. Dependencies should be tied to the packages directly and not meta-pacakges. If the dependencies are packaging only changes then we can fix them using overrides in amprolla. | 10:50 |
KatolaZ | Centurion_Dan: we can't make xfce4 depend on libpolkit-backend-consolekit | 10:50 |
KatolaZ | because people might want to use it without any session management | 10:51 |
KatolaZ | this is why you have task-*- packages | 10:51 |
KatolaZ | and this is definitely something we won't do for ASCII anyway | 10:51 |
KatolaZ | because that would require extensive testing | 10:51 |
KatolaZ | and would delay the release without any real reason | 10:52 |
Centurion_Dan | I often do installs without the task-* packages, because I dislike the selections. | 10:52 |
KatolaZ | great | 10:52 |
NewGnuGuy | Centurion_Dan: The devuanized version of dbus in ASCII is 1.10.22 which was the version in Stretch as of 9.2; however, in the current release of Stretch, 9.4, the dbus version is 1.10.26. | 10:52 |
KatolaZ | that should always be possible | 10:52 |
KatolaZ | and users should not take everything and the kitchen sink by default | 10:52 |
KatolaZ | if they want the kitchen sink, they go for the task-*-desktop package | 10:53 |
KatolaZ | we can't provide as many alternatives as users out there :) | 10:53 |
NewGnuGuy | https://tracker.debian.org/pkg/dbus | 10:53 |
KatolaZ | simply because we don't have enough people actually working on packages | 10:54 |
NewGnuGuy | Centurion_Dan: Should the newer dbus from Stretch be devuanized? | 10:54 |
KatolaZ | gnu_srs1: https://ci.devuan.org/job/policykit-1-binaries/architecture=all,label=all/18/console | 10:55 |
KatolaZ | gnu_srs1: the directory does not exist | 10:55 |
KatolaZ | you should create it first | 10:56 |
KatolaZ | (?) | 10:56 |
Centurion_Dan | NewGnuGuy: probably. I didn't package the latest dbus version for as far as I know... or it's not been updated in a long time... | 10:57 |
KatolaZ | NewGnuGuy: probably not for ASCII | 11:01 |
Centurion_Dan | KatolaZ: as I said we can apply this sort of fix via overrides in amprolla - and a recommends would for normally non task-* installs solve the issue without enforcing the installation of a session manager. It's a simple enough thing to do. | 11:01 |
KatolaZ | Centurion_Dan: which package should have that dependency? | 11:01 |
KatolaZ | it's not clear at all | 11:01 |
KatolaZ | (at least not clear to me) | 11:02 |
Centurion_Dan | I'm not 100% sure. I think it might actually be one of the login managers... probably slim.. | 11:02 |
parazyd | Centurion_Dan: What are amprolla overrides? | 11:02 |
KatolaZ | parazyd: maybe those are dak overrides? | 11:02 |
Centurion_Dan | parazyd: You should know, you would have implemented that in amprolla3... surely? | 11:03 |
NewGnuGuy | I was doing an unsupported offline install with the Devuan RC ISO combined with the first three discs of Stretch 9.4 and the newer, not-devuanized dbus from the Stretch disc caused the install to fail. | 11:03 |
KatolaZ | NewGnuGuy: you see | 11:03 |
parazyd | Centurion_Dan: Maybe. I didn't call anything an "override" so I'm unsure what you're talking about. | 11:03 |
parazyd | Hence the question. | 11:03 |
NewGnuGuy | Centurion_Dan: I have the install-log if you're interested. | 11:04 |
Centurion_Dan | parazyd: that's a bit of an oversight then... I'm surprised that there aren't broken jessie installs as a result.. | 11:04 |
parazyd | Dude | 11:05 |
parazyd | I asked a question | 11:05 |
parazyd | Can you answer? | 11:05 |
KatolaZ | Centurion_Dan: are you sure you are not referring to dak's overrides? | 11:05 |
NewGnuGuy | I'm guessing my unsupported install would have worked if I was using Stretch 9.2 discs, but I haven't tried it yet | 11:06 |
Centurion_Dan | KatolaZ, yes but it was implemented in Amprolla(original) too | 11:06 |
KatolaZ | uh? | 11:06 |
KatolaZ | what was implemented, exactly? | 11:06 |
gnu_srs1 | KatolaZ: Same problem as with xfce4-session. It is probably a build issue... | 11:09 |
KatolaZ | Centurion_Dan: if you are referring to getoverride in fs.py, it's not used anywhere in amprolla.... | 11:09 |
KatolaZ | no gnu_srs1 | 11:09 |
gnu_srs1 | bbl as gnu_srs | 11:09 |
gnu_srs1 | gotogo | 11:09 |
KatolaZ | ok | 11:09 |
Centurion_Dan | a feature that took a dak like overrides files to override package metadata in the packages files - mostly to change the section, but also occasionally to fix dependencies and add tags etc. It was a big part of amprolla and if amprolla3 didn't implement the same feature then we have lost a significant feature. | 11:10 |
KatolaZ | Centurion_Dan: read above^^^ | 11:10 |
KatolaZ | getoverride is not used anywhere in amprolla | 11:10 |
Centurion_Dan | Yes it was in fs.py. | 11:10 |
Centurion_Dan | It was!!! | 11:11 |
KatolaZ | how is this relevant to ASCII now? | 11:11 |
KatolaZ | sorry but I have real work to do | 11:11 |
KatolaZ | bbl | 11:11 |
KatolaZ | o/ | 11:12 |
Centurion_Dan | KatolaZ, because just like in jessie it would be an easy way to solve a problem that saves having to fork and/or rebuild packages | 11:13 |
Centurion_Dan | ... just for a metadata change in the repo. | 11:13 |
Centurion_Dan | parazyd: The overrides feature for amprolla allowed rewriting package metadata - section, dependencies, recommends etc based a per suite/section input obtained from http://packages.devuan.org/devuan/indices/ (something I note pkgmaster doesn't replicate) | 11:21 |
Centurion_Dan | as KatolaZ noted, it was part of the implementation of fs.py in original amprolla: | 11:26 |
Centurion_Dan | parazyd: see classes OverrideFS, IndexFS - dealt with getting the override files and in SuiteFS | 11:31 |
Centurion_Dan | and the key part is in lines 602 - 616 of fs.py, in the merge_dists function where it replaces the package metadata with any overrides that apply to that package. | 11:33 |
Centurion_Dan | parazyd: I did cover that in my description in a private irc conversation to you of amprolla's function way back when you were looking at doing the reimplementation that became amprolla3. | 11:41 |
Centurion_Dan | back in 27/03/17 | 11:42 |
parazyd | OK | 12:17 |
parazyd | Send a patch | 12:17 |
Centurion_Dan | parazyd: you make that sound so easy ;-) | 13:13 |
Centurion_Dan | I'll take a look at amprolla3. | 13:13 |
Centurion_Dan | and come back to you if I need help in creating a patch. | 13:14 |
gnu_srs | KatolaZ: I did not change anything else than debian/{rules,changelog} for polkit+devuan2.10. The build of +devuan2.9 was successful for ascii-proposed :D | 14:33 |
gnu_srs | So there is definitely a buildd issue. or Jenkins? | 14:34 |
gnu_srs | I'm AFK next 7 days, so if you need my attention you have to reply _tonight_ | 14:35 |
Ryushin | Can someone be so kind to let me know about the last comment on: https://github.com/zfsonlinux/zfs/issues/7540 "Is Devuan reversing Debian's decision to divide build and source? I think the logic I just pushed should be good enough, but you're going to need to retest it on all versions." | 14:58 |
Ryushin | I would think Devuan would be trying to stay as close to Debian as it can, ..... minus the systemd bits and other things that make better sense. | 14:59 |
KatolaZ | Ryushin: what are you talking about? | 15:18 |
KatolaZ | we are not touching zfs packages | 15:18 |
KatolaZ | unless there is a proper maintainer for that | 15:18 |
KatolaZ | which we don't have atm | 15:18 |
Ryushin | KatolaZ: They are talking about the build and source. Not necessarily zfs. | 15:49 |
Ryushin | I've been working with Debian to get the patches submitted which seems to go into a black hole. I'll soon work with the ZoL folks themselves to get a debian tree in the source. | 15:51 |
Ryushin | So no worries. | 15:51 |
KatolaZ | Ryushin: sorry but I don't understand :\ | 15:54 |
gnu_srs | (17:09:29) srs: KatolaZ: I just pushed a new version of polkit | 17:09 |
gnu_srs | Should fix the build issue. | 17:09 |
KatolaZ | ok gnu_srs | 17:10 |
KatolaZ | will have a look as soon as I can | 17:10 |
KatolaZ | thanks | 17:10 |
gnu_srs | KatolaZ: Can you please build polkit tonight? Otherwise I cannot help you ;) | 20:56 |
jaromil | KatolaZ: I think ryushin mentions he tried to push his patches to debian with no response, now will try "Zfs on Linux" which is a popular third-party report providing zfs to gnu/linux users | 21:08 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!