| teward | just to confirm: a Pre-Depends is weaker than a Depends, and if a Depends exists then it should already implicitly include a Pre-Depends because it has a *hard* Depends, right? | 01:09 |
|---|---|---|
| teward | (getting Lintian errors when putting both a Pre-Depends and a Depends on the same) | 01:09 |
| rbasak | teward: no, the inverse I think? | 01:20 |
| rbasak | https://www.debian.org/doc/debian-policy/ch-relationships.html#binary-dependencies-depends-recommends-suggests-enhances-pre-depends | 01:21 |
| teward | rbasak: hmm. so Pre-Depends is stronger than Depends then? So if I put a Pre-Depends i should drop the correspodning Depends? Getting a lintian fail of redundant-installation-prerequisite on this one package | 01:21 |
| rbasak | Pre-Depends is like Depends but additionally applies _before_ unpack, and circular Pre-Depends cannot be broken. | 01:22 |
| rbasak | Right | 01:22 |
| teward | ah okay, then i'll go poke that. | 01:22 |
| rbasak | But needing Pre-Depends is rare. | 01:22 |
| rbasak | Are you sure it's the right solution? | 01:22 |
| teward | rbasak: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1082373 (chasing this because we think we found a case where someone did something weird with policy and invoke-rc.d (https://launchpad.net/bugs/2081308) and it's manifesting weirdly) | 01:24 |
| -ubottu:#ubuntu-devel- Debian bug 1082373 in nginx "nginx: nginx missing pre-depends on nginx-common" [Important, Open] | 01:24 | |
| -ubottu:#ubuntu-devel- Launchpad bug 2081308 in nginx (Ubuntu) "open() '/etc/nginx/nginx.conf' failed (2: No such file or directory)" [Undecided, Incomplete] | 01:24 | |
| teward | rbasak: i agree, it may not be needed, but nginx pre-depending on nginx-conf makes sense though at the same time I don't like it. | 01:24 |
| teward | basically i'm trying to see if the Pre-Depends makes sense to (1) fix Debian bug and (2) the corresponding Ubuntu bug where someone fusses with policy and breaks things. | 01:30 |
| rbasak | teward: it's because there's a depends loop | 01:32 |
| rbasak | teward: why is there a depends loop? | 01:32 |
| teward | rbasak: because someone in the Debian team screwed up | 01:32 |
| teward | I'm pushing to fix it now | 01:32 |
| teward | not to Debian but to Salsa | 01:32 |
| rbasak | teward: I'd expect nginx to depend on nginx-common, and nginx-common not to depend on nginx, and then it'd work. | 01:32 |
| rbasak | I don't think a Pre-Depends makes sense here. | 01:33 |
| teward | i agree | 01:34 |
| teward | i have a headache staring at code | 01:34 |
| teward | rbasak: thanks for spot-checking me :) | 01:37 |
| teward | rbasak: so, nginx Depends: nginx-common. and remove the nginx-common Depends: nginx and that should resolve the loop? | 01:38 |
| rbasak | That's what I'd expect to see. But why did nginx-common depend on nginx? Is there a deeper flaw here that needs correcting. | 01:40 |
| rbasak | It's also odd (but not necessarily wrong) for nginx.conf to be in a different package to the package that ships the service. Why is this split needed. | 01:41 |
| rbasak | Apparently I'm tired enough that my questions are becoming statements. Time for bed :) | 01:42 |
| teward | rbasak: i don't know i didn't do that one :P I already put into Salsa the removal of the dep-chain because i don't think nginx-common should depend on nginx | 04:03 |
| teward | but i digress, Salsa for -3 has a pending change that removes the dep-chain by removing the nginx-common depending on nginx | 04:03 |
| teward | i'm headed to sleep too now | 04:03 |
| === gevfynezckwvfjst is now known as gljebhniabzzkuxn | ||
| === gljebhniabzzkuxn is now known as georgiag | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!