Story #5071
closedIssue #5063: Module defaults need to be considered and copied automatically during recursive dependency solving
As a user, a default module stream is copied together with its module_defaults
0%
Description
Motivation¶
If a module stream is marked as a default one, then nonmodular RPMs are allowed to depend on it.
If we copy the module without copying the module_defaults metadata, clients will not understand that it is special, and they will not be able to use that module to satisfy their dependencies, and the packages they provide will not be able to be installed normally through DNF as intended.
Solution¶
At copy time, always copy module_defaults if the stream specified in it is copied as well.
*How to merge module_defaults should be decided on this ticket before grooming
Updated by ttereshc over 5 years ago
- Related to Issue #5063: Module defaults need to be considered and copied automatically during recursive dependency solving added
Updated by dalley over 5 years ago
- Status changed from NEW to POST
- Assignee set to dalley
- Sprint set to Sprint 56
https://github.com/pulp/pulp_rpm/pull/1400
Adding as a subtask of parent issue https://pulp.plan.io/issues/5063
Added by dalley over 5 years ago
Added by dalley over 5 years ago
Revision c4cb08fb | View on GitHub
Remove 'obsoletes'
This is unnecessary, according to Igor.
Added by dalley over 5 years ago
Revision f52b0f2f | View on GitHub
Make changes to module default setup code
Change the name of module-default solvables to follow convention. Make sure module-defaults without a default stream aren't setting streams as defaults. Change the way provides are set up, as there should be no such thing as "module-default(name:stream)" since only one module-default exists per-module.
Added by dalley over 5 years ago
Revision a0a10884 | View on GitHub
Hack copy_rpms() to do more than just copy RPMs
Allow it to copy modules also. Ideally, we would do more refactoring to make this less hacky, but unless we change the scope of my work I would prefer to change as little as possible in that regard.
Added by dalley over 5 years ago
Revision 39e310ab | View on GitHub
Rewrite the recursive-conservative solver
Added by dalley over 5 years ago
Revision b474b613 | View on GitHub
Don't try to reuse Solver
It's supposed to be possible, but it's problematic in practice.
Added by dalley over 5 years ago
Revision c7c68551 | View on GitHub
Use empty strings for EVRs
I think Igor was wrong there, empty strings are fine.
Added by dalley over 5 years ago
Revision a2380f6b | View on GitHub
Revise note based on PR feedback
Updated by dalley over 5 years ago
- Status changed from POST to MODIFIED
- Platform Release set to 2.21.0
Updated by dalley about 5 years ago
- Status changed from MODIFIED to CLOSED - CURRENTRELEASE
Use placeholder EVR and Arch values
Libsolv does need these, so we have to provide fake ones, not just an empty string.
re: #5071 https://pulp.plan.io/issues/5071