1 |
> On 3 Nov 2021, at 22:15, Joshua Kinard <kumba@g.o> wrote: |
2 |
> Actually, it is possible to manage dependency errors like those. It just |
3 |
> takes a *lot* of elbow grease, and and long, long time time. Especially if |
4 |
> you have museum-grade hardware that these errors are happening on. |
5 |
> |
6 |
> For Perl, I've usually just uninstall everything under virtual/* first, then |
7 |
> try to let it upgrade. Sometimes that "unsticks" something in perl-core |
8 |
> enough to let the upgrades apply, pulling back in any needed items from |
9 |
> virtual/. If that doesn't solve the problem enough to let emerge do an |
10 |
> upgrade cycle, I'll try using just the @system target, or start yanking |
11 |
> things out from perl-core/* one-by-one until emerge shuts up and does what |
12 |
> it is told. |
13 |
|
14 |
For what it's worth, you really really shouldn't need to do this. |
15 |
|
16 |
Perl is great at consuming backtracking cycles (largely because |
17 |
of all of the || ( ... .... ) deps in the virtuals) and cranking that |
18 |
up helps a lot. |
19 |
|
20 |
But something which we're not really clear enough on is that |
21 |
users should genuinely never have to use emerge -C / --unmerge. |
22 |
|
23 |
Trying just @system shouldn't be needed either and is in fact |
24 |
likely to be more problematic: |
25 |
|
26 |
https://wiki.gentoo.org/wiki/Project:Portage/FAQ#Why_is_there_a_dependency_conflict_when_I_attempt_to_upgrade_a_single_package.3F |
27 |
|
28 |
> Also, *always* check for libperl-www being in the package list. It's |
29 |
> usually sucked in by way of dev-util/intltool and is responsible for ~35-40 |
30 |
> perl packages alone being pulled in. If that's in the list, try |
31 |
> uninstalling just that one, then run a depclean to remove all of its |
32 |
> dependencies and then see if the upgrade will work. If the upgrade tries to |
33 |
> drag intltool or libperl-www back in, use --exclude to hold it out for later. |
34 |
> |
35 |
|
36 |
Aye, we fixed eudev to not need it anymore and there's an avahi |
37 |
bug for the same I'll look at shortly. |
38 |
|
39 |
> That all said, am I alone in thinking that the way Portage emits error |
40 |
> messages about dependency resolution problems is extremely messy and |
41 |
> border-line unreadable at times? The current way it outputs depgraph errors |
42 |
> feels like something I'd expect from a --debug switch. We've got a |
43 |
> reputation for being playful and colorful on the command line with our |
44 |
> tooling, so I would wonder if that depgraph output couldn't be made to |
45 |
> look....nicer? |
46 |
> |
47 |
|
48 |
Yeah, I think one of our biggest weaknesses is that there's |
49 |
usually a LOT of output and figuring out what matters isn't easy. |
50 |
|
51 |
A good rule of thumb is that the "most fatal" error is _usually_ |
52 |
at the bottom but this isn't always true. |
53 |
|
54 |
Best, |
55 |
sam |