lp:~apw/ubuntu-archive-tools/copy-proposed-kernel--support-lrg-lrs

Branch merges

Propose for merging

No branches dependent on this one.

Merged into lp:ubuntu-archive-tools at revision 1462

Łukasz Zemczak: Approve on 2021-03-11

Diff: 533 lines (+239/-72)

1 file modified

copy-proposed-kernel (+239/-72)

api

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Related blueprints

1464. By Andy Whitcroft on 2021-03-11

copy-proposed-kernel: clean up whitespace

Clean up whitespace based on review feedback.

Signed-off-by: Andy Whitcroft <email address hidden>

1463. By Andy Whitcroft on 2021-03-11

copy-proposed-kernel: commonise error handling via RoutingError

Switch remaining direct error/exits in routing to raising RoutingError's.
This will allow those to be tested sensibly.

Signed-off-by: Andy Whitcroft <email address hidden>

1462. By Andy Whitcroft on 2021-03-10

copy-proposed-kernel: handle linux-restricted-generate and linux-restricted-signatures

Routing for linux-restricted-generate and linux-restricted-signatures
is non-standard. Specifically they LRG is embargoed and may only be
seen in a private build PPA (typically a -ps adjunct PPA) and LRS will be
in the same place. These will move to signing and then LRS (only) will
move on to proposed. Add support for the alternative build-private route
selecting this by default for LRG and LRS. Also add checks for copying
LRG to anywhere other than a signing PPA. Finally treat LRS in the same
manner as linux-signed so that builds and picks up signatures appropriatly.
Add new tests for all of this.

Signed-off-by: Andy Whitcroft <email address hidden>

1461. By Andy Whitcroft on 2021-03-10

copy-proposed-kernel: iterate over all copies under dry-run

Signed-off-by: Andy Whitcroft <email address hidden>

1460. By Andy Whitcroft on 2021-03-10

copy-proposed-kernel: copies from signing/security to safe public locations need unembargo

Both security and signing PPAs are private and for some kernels we copy
to the primary archive -proposed pocket or other public PPA. We need to
supply unembargo to the copy in this case.

Though we could ask Launchpad if the destination is public, not only is
the horribly expensive it would be the equivalent of supplying unembargo
always rendering the protections it offers useless. Better that we
only request unembargo for combinations that we know to be safe. That
is from security and signing PPAs to known public PPAs.

Finally, we also have one anomolous kernel which in its early life had a
private build PPA but was published to -proposed; add a manual override
for this case.

Signed-off-by: Andy Whitcroft <email address hidden>

1459. By Andy Whitcroft on 2021-03-10

copy-proposed-kernel: handle up to three security PPAs

We now have three security PPA sets (ppa, ppa2, ppa3, devel, devel2,
devel3, etc). Add support for --security3 and refactor the internals
to use archive counts rather than individual flags. Update the self-tests
to use the new form.

Signed-off-by: Andy Whitcroft <email address hidden>

1458. By Andy Whitcroft on 2021-03-10

copy-proposed-kernel: include the series in copy detail output

We output the source and destination archive and pocket information in
our detail output but do not include the series. Add this.

Signed-off-by: Andy Whitcroft <email address hidden>

1457. By Andy Whitcroft on 2021-03-10

copy-proposed-kernel: detect signed packages by their type

We currently detect signed packages by name comparisons better to use
the package type information.

Signed-off-by: Andy Whitcroft <email address hidden>

1456. By Iain Lane on 2021-03-08

merge trunk

1455. By Iain Lane on 2021-03-04

promote-to-release: Use copyPackage's move support

Currently we copy a package from proposed to release, and then separately
delete it from proposed. Either the copy or the deletion can fail, and then we
can end up with incomplete migrations. It's better to have both or neither
happen.

`copyPackage()` on the Launchpad API has a `move` boolean parameter that
combines copying and deletion into one call. Use that instead. This doesn't
solve all cases where the copier might reject a copy we ask it to do, but it
does mean that at least in those cases the package will still be in proposed
rather than nowhere as could happen previously.