Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!

Bug 595224

Summary: sys-apps/portage: detect conflicts that appear to be solvable with 1 additional backtracking run, and solve them
Product: Portage Development Reporter: Yanestra <Yanestra>
Component: Core - DependenciesAssignee: Portage team <dev-portage>
Status: CONFIRMED ---    
Severity: normal CC: bkohler, devurandom, esigra, pacho
Priority: Normal    
Version: unspecified   
Hardware: All   
OS: Linux   
See Also: https://bugs.gentoo.org/show_bug.cgi?id=615680
Whiteboard:
Package list:
Runtime testing required: ---
Bug Depends on:    
Bug Blocks: 300071    

Description Yanestra 2016-09-26 18:21:54 UTC
With no parents demanding the older version, what is thr reason for the problem?

kde-plasma/kde-cli-tools:5

  (kde-plasma/kde-cli-tools-5.6.5:5/5::gentoo, ebuild scheduled for merge) pulled in by
    (no parents that aren't satisfied by other packages in this slot)

  (kde-plasma/kde-cli-tools-5.7.5:5/5::gentoo, installed) pulled in by
    >=kde-plasma/kde-cli-tools-5.7.5:5 required by (kde-plasma/kscreen-5.7.5:5/5::gentoo, installed)
    ^^                         ^^^^^^^
Comment 1 Ben Kohler gentoo-dev 2016-09-26 18:26:16 UTC
If your emerge isn't continuing, there is probably another more serious problem.  The output you showed is almost certainly a false-positive
Comment 2 Zac Medico gentoo-dev 2016-09-26 18:28:43 UTC
Were there any other conflicts in the same calculation? Generally, this type of behavior means that it ran out of backtracking runs before it settled on a valid solution. You can try using a larger --backtrack setting (default is 3).
Comment 3 Yanestra 2016-09-26 23:57:22 UTC
This was the only problem. Apparently, kde-plasma/kde-cli-tools-5.7.5 became flagged ~amd64 after installation. If I accept ~amd64 for that package, the problem disappears.

What I want to say is, the error message is quite enigmatic.
Comment 4 Zac Medico gentoo-dev 2016-09-27 00:01:57 UTC
Yeah, it's a common source of frustration for users. It's also accompanied by a message suggesting to use a larger --backtrack value, which is the correct course of action. If a larger --backtrack value does not solve it, then we should investigate that.
Comment 5 Yanestra 2016-09-27 01:29:09 UTC
Sigh. If that issue's known, I might as well close this bug.

A candidate for the wishlist, maybe.
Comment 6 Zac Medico gentoo-dev 2016-09-27 01:45:36 UTC
We could add some code to detect this specific case, where we have a slot conflict that appears to be solvable with 1 additional backtracking run, and allow for an extra backtracking run in this case.
Comment 7 Zac Medico gentoo-dev 2016-09-27 01:53:30 UTC
Also, I suspect that there may be a bug which prevents it from reaching the solution, even when given 1 additional backtracking run. It may be due to the resolver's preference for upgrades, since we want to upgrade packages whenever possible (and comment #0 shows a conflict which is solvable only by missing an upgrade).
Comment 8 Zac Medico gentoo-dev 2017-03-13 16:54:26 UTC
*** Bug 430190 has been marked as a duplicate of this bug. ***
Comment 9 Zac Medico gentoo-dev 2017-03-17 18:37:11 UTC
(In reply to Zac Medico from comment #6)
> We could add some code to detect this specific case, where we have a slot
> conflict that appears to be solvable with 1 additional backtracking run, and
> allow for an extra backtracking run in this case.

We could have a scoring system that is based on the number of problems with the dependency graph, and use that to decide if it's appropriate to automatically allocate some more backtracking.