libera/#devuan-dev/ Saturday, 2020-11-21

Unit193LeePen: There's a lot of noise in http://paste.openstack.org/show/R0Sup8WDxVRNRAzgYyFn, but B-D, S-V, Section, Priority, whitespace, and d/copyright would be noteworthy.  d/watch was only adjusted to simplify and match unreleased versions.00:00
bgstack15LeePen: is the installer-iso tool going to use beowulf-proposed-updates by default?00:11
bgstack15Or will it use it only if the system that builds it has that enabled?00:11
rrqbgstack15: TLDR; if the build host has that source point, the iso is likely to become "version confused" about those packages00:39
bgstack15I thought we were building a new iso, to get the proposed-updates category.00:40
rrqits first stage that sets up its meta database about packages will ignore that source list point00:40
rrqits second stage that pulls the packages will pull using that spurce point00:40
rrqI'd agree on the objecive :)00:41
rrqt00:41
rrqprobably will need pacthing of pool/Makefile to be aware of that source point00:42
bgstack15And I hate to quibble, but ensuring I know the vocabulary precisely: please tell me that you want my build host to include the binaries repository:00:42
bgstack15deb http://deb.devuan.org/merged beowulf-proposed-updates main contrib non-free00:42
bgstack15in the build host's sources.list00:42
rrqyeah I don't think anyone has put their mind to this question; and you sound stressed :)00:47
bgstack15Ah, you may relax. I'm not stressed. Mildly frustrated, but not stressed.00:47
bgstack15I intend to do the right thing, without too many failed attempts.00:47
rrqif it's to be a "proper" 3.1 iso, then it should not include a "side" codename beowulf-proposed-updates, but only the codename "beowulf"00:48
bgstack15interesting to note, that my brief perusal of the pool/Makefile leads me to believe that the presence of the list files in /var/lib/apt/lists/ will mean they will be parsed.00:48
bgstack15Of course, those files will appear if at any point apt-get update is run when an apt repo is enabled. If it is subsequently disabled, those files will still exist and become stale.00:49
rrqits limited by CODENAME and SECTIONS00:49
bgstack15OK, so it sounds like I should disable beowulf-proposed-updates (and clean apt's lists cache of it) on the build host.00:49
bgstack15This sounds however, like it will defeat LeePen's comment above about how tomcat9 is now in beowulf-proposed-updates.00:49
rrqyes00:50
rrqthere is that special patch to add beowulf-security to the iso; maybe this expands to a second one to also include beowulf-proposed-updates00:52
bgstack15And let's make me look even dumber for the irc logs! When I've run this before, I never saw a prompt about signing the ISO file. Will I sign the iso artifacts after their generation?00:52
bgstack15Forgive me; I'm new to actually participating in real distro work!00:52
golinuxYou are the next generation, grasshopper.  :D00:53
golinux( you might not get that reference)00:54
rrqthe signing of the published iso is a manual chore00:54
rrqI have had all sorts of learning experiences to get that done00:56
rrqso the technical issue is to work around the small-mindedness of the original packager to allow the packages of the additional codename to be included; either by pretending them to be in "main" (as is done for -security) or by allowing the isoe to carry an additional codename00:59
rrqisoe = iso00:59
golinux(Apologies for the OT) https://www.youtube.com/watch?v=gbNCBVzPYak01:00
rrqthe second approcahe which on the surface seems more proper will however find resistance in the acutal installation procedure, which currently, like debootstrap, assumes and relies on a single codename (beowulf)01:02
rrq(that was why beowulf-security/main is patched in as if being in "beowulf/main")01:04
rrqso it probably comes back to a patch of pool/Makefile to bring in the beowulf-proposed-updates/main packages as if being beowulf/main in its meta database, and then pull those packages onto the iso01:06
rrqI think it would as simple as to replicate the SECURITY patch in that Makefile01:07
rrqand include the sources.list point on the build host01:07
rrqlines 28,2901:07
bgstack15So I just dist-upgraded with the beowulf-proposed-updates, and now I have disabled that repo. But now I've got the ~9 packages, and apparently installer-iso wants to use the currently-installed list to help populate its list.01:15
bgstack15I am having trouble getting apt-get --allow-downgrades dist-upgrade to actually downgrade the ~9 packages...01:15
bgstack15And yes, I am familiar with the grasshopper reference. I saw that movie once.01:16
golinuxDidn't know it was a movie.  It was a TV series in the 70s01:17
rrqbiab01:18
bgstack15Oh, I actually thought it was from The Karate Kid.01:19
golinuxhttps://www.imdb.com/title/tt0068093/01:19
golinuxBullseye going into freeze soonish.  Posted by HoaS on FDN: http://forums.debian.net/viewtopic.php?f=19&t=14804901:24
rrqright: the iso should have the ~9 packagaes... the "problem" is to get those into the "meta database" during build so they get entered into the ISO's beowulf/meta Packages file02:05
rrqso use pool/Makefile:28 to define a PROPUP variable similar to SECURITY and then add that similarly o line 2902:06
rrq(which I suppose becomes line 30)02:07
rrqand then all magic will happen with just an ordinary incantation02:08
HumanG33kdebian motion package drop sysvint script in testing (i make a mail to ask to continue ship the script )20:36

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