Merge lp:~xnox/britney/oem-osp1 into lp:~ubuntu-release/britney/hints-ubuntu

Proposed by Dimitri John Ledkov
Status: Rejected
Rejected by: Steve Langasek
Proposed branch: lp:~xnox/britney/oem-osp1
Merge into: lp:~ubuntu-release/britney/hints-ubuntu
Diff against target: 10 lines (+1/-0)
1 file modified
ubuntu-release (+1/-0)
To merge this branch: bzr merge lp:~xnox/britney/oem-osp1
Reviewer Review Type Date Requested Status
Steve Langasek Disapprove
Andy Whitcroft Pending
Ubuntu Release Team Pending
Review via email: mp+371570@code.launchpad.net
To post a comment you must log in.
Revision history for this message
Steve Langasek (vorlon) wrote :

Does the original rationale for badtesting all of the custom kernels still apply? Looking at http://autopkgtest.ubuntu.com/packages/l/linux-oem-osp1/eoan/amd64 I see a successful rebuild test against glibc; a number of 'neutral' test results for testing the actual kernel (as expected since the autopkgtest instances won't actually boot any kernel except generic); and then a number of legitimate rebuild test failures due to the fact that this kernel still needs updated to 5.2 in order to fix the compatibility problem with the current eoan toolchain.

So why is an 'all' hint in order if the current failure is transient and package-version-specific?

review: Needs Information
Revision history for this message
Dimitri John Ledkov (xnox) wrote :

A binary copy package from the bionic pocket to higher series is only required to be buildable with the bionic toolchain and no other.

That v5.0-based kernel flavour will never be meant to be compiled with a non-bionic toolchain, as a different series will be in use if and when osp2 comes out.

Since the package has never been uploaded to eoan series, and is not intended to be uploaded to eoan, all hint is the only one that makes sense.

This is not dissimilar from any other bionic-only copy-up flavours like linux-oem

Revision history for this message
Dimitri John Ledkov (xnox) wrote :

oem kernels are sometimes based on higher version number, thus by accident, they were buildable once they were matching the major version of the release in question (i.e. v5.0 == disco linux-generic), but it has never been intentional for that to be the case.

Revision history for this message
Dimitri John Ledkov (xnox) wrote :

Please create ubuntu-hints project
Please enable bug tracker

Discussing "please unblock zlib because of bogus linux-oem-osp1 regression" is not productive over a bzr merge proposal of a hint text, and how it should be done, it's better suited as a bug against "ubuntu-hints" project, cause then one can pile on extra tasks against series as needed.

Revision history for this message
Steve Langasek (vorlon) wrote :

target repo moved to git; please resubmit there if appropriate.

review: Disapprove

Unmerged revisions

3813. By Dimitri John Ledkov

Add osp1 variant to the oem hint.

Preview Diff

[H/L] Next/Prev Comment, [J/K] Next/Prev File, [N/P] Next/Prev Hunk
1=== modified file 'ubuntu-release'
2--- ubuntu-release 2019-08-20 20:35:51 +0000
3+++ ubuntu-release 2019-08-21 11:58:29 +0000
4@@ -172,6 +172,7 @@
5 force-badtest linux-gcp/all
6 force-badtest linux-gke/all
7 force-badtest linux-oem/all
8+force-badtest linux-oem-osp1/all
9 # fails rebuild test due to timeouts, but linux/arm64 passes
10 force-badtest linux-snapdragon/all
11