libera/#devuan-dev/ Tuesday, 2018-10-30

IrrwahnKatolaZ: It looks like you just removed su from util-linux, but given the wording in util-linux NEWS file, wouldn't it make more sense to drop su from the login package (src:shadow)?14:40
IrrwahnKatolaZ: (Just asking, please disregard in case I'm not getting the whole picture here.)14:41
KatolaZIrrwahn: login is essential in sid15:06
KatolaZmaybe I am missing something15:06
KatolaZthe util-linux package in sid does not ship su any longer15:07
KatolaZsince it is provided by login (shadow)15:07
IrrwahnOh, that's interesting, given the wording in util-linux.NEWS for version 2.32-0.4, I quote:15:08
Irrwahn"The util-linux implementation of /bin/su is now used, replacing the one previously supplied by src:shadow (shipped in login package), and bringing Debian in line with other modern distributions. "15:08
KatolaZoh FFS15:09
IrrwahnIt goes on to describe the differences between the two implementations, which do not seem major, but significant.15:09
KatolaZsorry Irrwhan15:09
KatolaZgone through that between a lecture and a meeting15:09
KatolaZ:\15:09
IrrwahnKatolaZ: No offense taken (why would I?), so no need to apologize. :D15:10
KatolaZlet me see if debootsrap works, first15:10
KatolaZwell, this created confusion, and I am the source of it :)15:11
IrrwahnDoing the same here right now, pulling from pkmaster.15:11
IrrwahnShift happens. ;o)15:11
KatolaZwe should update sysvinit as well15:13
KatolaZ'cause the version in ceres is too old15:13
IrrwahnSounds reasonable.15:13
IrrwahnYep, debootstrap falls flat on configuring sysvinit-utils.  :/15:15
KatolaZyep15:15
KatolaZworking on that15:15
IrrwahnGreat, thanks a bunch! :)15:16
KatolaZbuilding now the same version as in beowulf15:19
KatolaZI will update them as soon as the new sysvinit pkg gets in debian15:20
IrrwahnYup, the beowulf one should work for the moment I assume.15:21
KatolaZthere is a problem with one of the cows anyway15:21
IrrwahnMeaning one of the the build systems? (sorry if that's a stupid thing to ask)15:22
KatolaZyep15:23
KatolaZit's building for amd64 though15:23
KatolaZlet's see if it works first15:23
KatolaZand then we will fix it15:23
Irrwahnsounds like a plan15:24
KatolaZok, finishing now I guess15:25
KatolaZit should hit the repos within a couple of minutes15:26
Irrwahncool, will re-run the debootstrap test as soon as it does15:28
Irrwahndpkg: dependency problems prevent configuration of sysvinit-utils:15:37
Irrwahn util-linux (2.32.1-0.1+devuan2) breaks sysvinit-utils (<< 2.88dsf-59.4~) and is unpacked but not configured.15:37
Irrwahn  Version of sysvinit-utils to be configured is 2.88dsf-59.3+devuan2.15:37
IrrwahnKatolaZ: ^^^^^^^^15:37
KatolaZhold on15:38
KatolaZare you using pkgmaster?15:38
Irrwahnyep15:38
KatolaZit has just finished syncing15:38
Irrwahnoh, maybe I was to quick on the trigger15:38
KatolaZtrying a debootstrap here as well15:39
KatolaZsysvinit was installed15:39
KatolaZdebootstrap continues...15:39
KatolaZnope15:40
KatolaZ:D15:40
KatolaZthe problem is in udev now15:40
KatolaZbut this is in base-system15:40
KatolaZso it should keep going I guess15:40
KatolaZ(but we need to fix that anyway)15:41
IrrwahnISTR aitor wrote something about udev/eudev in the DNG thread.15:41
KatolaZhaven't had time to read15:42
Irrwahnshould be upthread to our exchange on DNG15:42
KatolaZwill have a look later I guess15:42
KatolaZwe just need the udev dummy package15:43
KatolaZalso, we should bump eudev in unstable15:43
IrrwahnYep, aitor wrote in DNG, I quote: "An empty dummy package for udev in Ceres is required too. Otherwise, debootstrap will not work: libudev1.so exists in both udev and eudev."15:44
IrrwahnSo ... add udev dummy; bump eudev; drop su from (shadow)login; revert drop su from util-linux ... that should be it, right?!15:46
Irrwahn(in theory, at least)15:46
Irrwahnactually: bump (shadow)login to (>= 1:4.5-1.1~)15:49
IrrwahnThat should be the version w/o su I would guess, as it is referenced by util-linux.15:50
KatolaZamesser:21:05
amesserHi Katolaz! I'm sorry, unfortunately I did not yet had time to discuss the 'percolate' thing21:10
amesserbut, this evening, I would be available21:11
amesserso may I ask now some questions?21:12
KatolaZplease do21:20
KatolaZI will probably not be at the meeting21:20
amesserok21:20
amesserI remember, a year ago or so, you said, packaging after ascii will be different: "Versions will percolate down from unstable to stable".21:21
amesserwhat does this mean for a git repository21:22
amesserassuming I update to latest upstream version, this typically goes to branch "master"21:22
amesserin changelog, do i put "unstable" in there?21:23
KatolaZit means that from now on we will onlu have one active branch21:23
KatolaZsuites/unstable21:23
KatolaZand we will build only for unstable21:23
amesserok21:23
amesserany new version just gets iunto this branch then21:24
KatolaZupstream should go to an 'upstream' branch21:24
amesserjep21:24
KatolaZand should be merged into suites/unstable as approproate21:24
amesserok21:24
KatolaZobviously, you could have more WIP branches if needed21:24
KatolaZ(well, as needed)21:24
KatolaZamesser: does it make sense?21:25
amesserIn changelog of suites/unstable the entries for the new version then always get "unstable", right?21:25
KatolaZyou mean the debian/changelog?21:26
amesseryes21:26
Centurion_DanKatolaZ: building only for suites unstable is wrong approach in my opinion, because migrations won't be able to detect breakages.21:26
KatolaZuh?21:26
KatolaZmigrations to/from what?21:26
KatolaZunstable will be basically aligned with beowulf21:26
KatolaZ(beowulf might be a little behind)21:26
KatolaZ(but just a little)21:26
amesserassuming a new version added to suites/unstable has been there for a while and is now considered stable21:27
KatolaZnamely21:27
amesserhow will it go on to "suites/stable" ?21:27
KatolaZit won't21:27
Centurion_DanI agree we build for unstable first, but we should always build for testing as the migration process - its the only way we will pickup potential breakages due to dependencies dak can't21:27
KatolaZwell, it's a waste21:28
KatolaZif beowulf is aligned with unstable21:28
KatolaZbreakages must be resolved in unstable21:28
KatolaZboth for unstable and testing21:28
Centurion_Danand doesn't know about all dependencies because it's only got a partial view of all packages21:28
KatolaZno Centurion_Dan21:28
KatolaZforget the current situation21:28
KatolaZwhere unstable and testing are not aligne21:28
KatolaZtesting must be a non-broken version of unstable21:29
KatolaZalways in sync21:29
KatolaZexcept for stuff that could break testing21:29
amesserok, lets talk about esting21:29
amessertesting21:29
KatolaZif there is a breakage, we fix it in instable21:29
KatolaZ~unstable21:29
amesseris tresting also a branch "suites/trsting" ?21:29
KatolaZI guess it won't be necessary21:29
amesserhmm21:30
KatolaZif we only build for unstable21:30
Centurion_Danbut transitions like the util-linux/shadow one will always silently cause breakages for us unless we are perfectly aligned timing wise with debian.21:30
KatolaZand we tag the relevant versions21:30
KatolaZwell, that's a specific case21:30
KatolaZ:)21:30
KatolaZwhich will cause breakage anyway21:30
amesserah, you mean we just build suites/unstable21:30
KatolaZyes amesser21:30
KatolaZthat is my proposal21:30
KatolaZas in Debian21:31
KatolaZwe just build suites/unstable21:31
KatolaZfor unstable21:31
amesserand if we reach a version which fits, it is tagged?21:31
KatolaZthe stuff that percolates to testing is moved to testing21:31
Centurion_DanOur dak doesn't know about all dependency issues because it doesn't know about all the packages in the merged archive.21:31
KatolaZCenturion_Dan: but aptly might know :)21:31
KatolaZanyway, let's try it21:31
KatolaZif it does not work, we can always maintain a suites/testing21:32
KatolaZand build for testing21:32
KatolaZwhen it's needed21:32
Centurion_DanIt may work once we have aptly setup.21:32
KatolaZbut only if it's needed21:32
KatolaZyep Centurion_Dan21:32
KatolaZthat's the plan21:32
KatolaZamesser, Centurion_Dan: but let's think about it and discuss it21:32
amesserso percolating is a move of the already build package within the package repos, but the git repo is not affected21:32
KatolaZI would suggest to start building for unstable21:32
KatolaZif it becomes impractical, we will see21:33
Centurion_DanWe can use dak or aptly to kick off the build process though - and as part of that do external dependency checks.21:33
KatolaZyes amesser21:33
KatolaZthat's the idea21:33
KatolaZand you have all the built versions tagged21:33
amesserok, now i got it21:33
amessertha makes perfectly sense for me21:33
KatolaZsorry guys21:33
KatolaZI won't be at the meeting21:34
KatolaZI am wasted tited21:34
KatolaZ~tired21:34
KatolaZCenturion_Dan: util-linux is building21:34
Centurion_DanKatolaZ: I agree that in principal nothing should be built for testing that hasn't been in unstable - down to the version.21:34
Centurion_DanCool.21:34
amessernext meeting is tomorrow, right?21:35
KatolaZoh what?21:35
KatolaZisn't today Wednesday?21:35
KatolaZo_O21:35
KatolaZoh FFS21:35
amesserlol21:35
KatolaZI am an idiot21:35
KatolaZ:D21:35
KatolaZ(a tired one, but still...)21:36
KatolaZ:D21:36
KatolaZok see you around21:36
KatolaZsorry21:36
KatolaZo/21:36
onefangIt's Wednesday on this side of the planet.  Get some rest KatolaZ,21:37
amessersince we'Re no back to  standard time, i think i can attend tomorrow as well21:37
golinux"<KatolaZ> I won't be at the meeting"22:15
golinuxKatolaZ: The meet is tomorrow.22:15
golinuxHere's the pad BTW: https://pad.dyne.org/code/#/1/edit/cfbrjT5XEVF44vUC4R+SFA/EZ0J4iLANNc5Hajtx8Yd+zUs/22:16

Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!