ACK: [SRU v3][J:lrm-hwe-5.19/J:lrm-hwe-6.2/M:lrm/U:lrm][PATCH 0/1] Add lrm autogenerated transitional independent of variants

[Andrea Righi]

On Thu, May 11, 2023 at 01:54:14PM +0300, Timo Aaltonen wrote:
> From: Timo Aaltonen <timo.aaltonen at canonical.com>
> 
> BugLink: https://bugs.launchpad.net/bugs/1991130
> 
> v3: Add <version> so that NNN-server and -open get handled correctly, and use
> the same loop as for default variant transitionals
> 
> [Impact]
> 
> The kernel lrm packages have a mechanism to auto-generate transitional packages
> by adding a line to 'debian/package.config' in the following format
> 
> transitional <version> <from> <to> <archs>
> (note: <version> is new from this patch)
> 
> Example:
> 
> transitional 450-server oem-22.04a generic amd64
> 
> When being processed by 'debian/scripts/generate-rules' the autogenerated
> contents are added to 'debian/control.d/transitionals-local'. However, this file
> is not included in the final 'debian/control' for packages which does not
> contain the default variant '--' (see debian/rules.in). This is intentional to
> prevent unwanted transitionals.
> 
> However, there are some cases where we want to transition packages to kernels
> which are not the default variant in a series (e.g. hwe kernels).
> 
> [Fix]
> 
> The proposed fix is to add a new command "migrate", which does exactly what
> "transitional" does but adding the content to 'debian/control.d/migrate-local'
> which would get added to the final 'debian/control' independently of the
> source's variants.
> 
> [Test Case]
> 
> Add "migrate" entries to 'debian/package.config' and check whether they get
> added to 'debian/control' even for non default variant packages.
> 
> [Where problems could occur]
> 
> If not used correctly, unwanted transitionals could be generated causing issues
> to the archive. However, this is a new command and should not impact any of the
> previously set transitionals and it needs to be explicitly used.

I don't understand all the details of this, but it seems that adding the
new command "migrate" doesn't affect the previous behavior of
"transitional", so it looks safe to me.

Therefore:

Acked-by: Andrea Righi <andrea.righi at canonical.com>