You are not logged in.
Pages: 1
I am starting to use Arch a bit, after getting into a couple of annoying problems with Debian. I went to install nvidia, and I found a few dependency cycles.
So I did some research, and I seem to get one basic answer: don't worry about it.
Is it normal? I mean, I have dependency cycles and I just neglect them, just because everything works? There is no hope of having these issues solved?
I know that unfortunately this kind of problem can occur, but the idea of considering it a non problem doesn't make me very happy.
Anybody can give me more information?
Thank you.
Offline
Dependency cycles happen. If they don't cause problems, why are you worried about them? Sometimes they're valid.
Offline
Dependency cycles happen. If they don't cause problems, why are you worried about them? Sometimes they're valid.
Because I might have a problem without knowing it; maybe I will find out next month. And I might spend a few hours trying to solve it.
Or maybe someone else will have a problem.
In general, dependency cycles are not a good idea. Installing software with an unmet dependency, even if maybe just during installation, makes me feel a little uncomfortable.
Offline
Theoretical problems are not really problems. While dependency cycles are nice to avoid, they aren't an actual problem or they wouldn't be put into the repos that way.
Offline
If you are really worried, you can install the cycle twice. Then the dependency is met at install time too. I wonder if pacman should do that...
Offline
If you are really worried, you can install the cycle twice. Then the dependency is met at install time too. I wonder if pacman should do that...
Not as much of a problem now with hook support, most of that stuff is being run after everything is updated.
Offline
Theoretical problems are not really problems.
Until they turn practical :-)
While dependency cycles are nice to avoid, they aren't an actual problem or they wouldn't be put into the repos that way.
Okay so that means that I should disregard every dependency cycle I see?
On the other side, if dependency cycles are nice to avoid, it might help to solve them :-)
Not as much of a problem now with hook support, most of that stuff is being run after everything is updated.
you mean that the dependency cycle problem is actually solved? Kind of, at least?
Offline
It used to be a rather common occurrence, but the only one i've seen recently is harfbuzz / icu I think.
Changes in pacman have reduced those messages a lot.
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Online
Dependency cycles are warnings, not errors, due to their not being errors.
And the issue is entirely due to the fact that they introduce a case *mid-transaction* where one thing is installed but its dependency is not.
It doesn't actually matter, unless something tries to execute the package, again, mid-transaction. as Scimmia said, we've moved to using PostTransaction hooks for many things, so having install scripts run commands mid-transaction and failing because of dependency cycles is... less likely now.
If anything actually breaks because of dependency cycles, that would go from a *warning* to an *error*, and is something actionable we could actually do something about...
Managing AUR repos The Right Way -- aurpublish (now a standalone tool)
Offline
It doesn't actually matter, unless something tries to execute the package, again, mid-transaction. as Scimmia said, we've moved to using PostTransaction hooks for many things, so having install scripts run commands mid-transaction and failing because of dependency cycles is... less likely now.
Thanks. Not exactly what I was hoping for, but it does help a little.
Offline
Pages: 1