Copy from PPA with binaries evades NEW and puts new packages in their default component
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| | Launchpad itself |
High
|
Colin Watson | ||
Bug Description
We regularly copy kernels from the kernel team PPA to -proposed. Whenever there is an ABI bump, i. e. the source packages build binary packages with a new name, these new packages are technically binary NEW for Ubuntu. But the syncs generated by
ubuntu.
only land in unapproved, and all binaries are automatically accepted into universe. It would be better if these syncs would land in NEW, from where any archive admin or SRU team member could override them to main. Right now we require a person who can ssh to cocoplum to run change-override.py to move the binaries from universe to main.
Related branches
- William Grant: Approve (code) on 2014-05-30
-
Diff: 146 lines (+67/-8)3 files modifiedlib/lp/soyuz/configure.zcml (+1/-2)
lib/lp/soyuz/model/packagecopyjob.py (+27/-6)
lib/lp/soyuz/tests/test_packagecopyjob.py (+39/-0)
| Colin Watson (cjwatson) wrote : | #1 |
| Changed in launchpad: | |
| status: | New → Triaged |
| importance: | Undecided → High |
| tags: | added: derivation |
| tags: | added: package-copies |
| Changed in launchpad: | |
| status: | Triaged → In Progress |
| assignee: | nobody → Colin Watson (cjwatson) |
| tags: | added: package-overrides |
| Adam Conrad (adconrad) wrote : | #2 |
The "and puts packages in universe" part of this bug has been more or less fixed with binaries now defaulting to the component of their source. That said, copies bypassing binary NEW is still a bug.
| summary: |
- Copy from PPA with binaries evades NEW and puts new packages into - universe + Copy from PPA with binaries evades NEW and puts new packages in their + default component |
| Dmitry Shachnev (mitya57) wrote : | #3 |
There is also an issue when in PPA, you can build a package from main against a package from universe, then binary copy both to archive, and both will be accepted.
A recent example is bug 1428162, which was noticed after the package was uploaded and accepted.

I agree - I think NEWness should be determined relative to the archive you're copying into.