P-A-S support needs proper binary-only excludes

Bug #30264 reported by Daniel Silverstone
4
Affects Status Importance Assigned to Milestone
Launchpad itself
Fix Released
Critical
Christian Reis

Bug Description

For now the buildd queue builder ignores P-a-s lines which do not start with a % symbol.

This means that we sometimes try and build something when we shouldn't, but currently the data model only allows us to never build something we should.

Essentially we need to trust the Binaries line of the dsc but we don't currently expose that in the data model.

Revision history for this message
James Troup (elmo) wrote :

Boring background: quinn-diff follows a 'must have everything possible built to be considered up-to-date', wanna-build and launchpad buildd follows the 'must have _something_ built to be considered up-to-date'. This makes supporting P-a-s' binary only entries a little more interesting.

Essentially, I think the algorithm when determining up-to-dateness needs to be something like:

 If a sourcepackagerelease has more than one package to build and there is a binarypackagerelease associated with this distroarchrelease:
   consider it up to date
 otherwise, if a sourcepackage release has only one package to build and that binary package is listed in P-a-s in such way that excludes that binary from building for this distroarchrelease:
  consider it up to date

The problem here, as Daniel says, is determining the 'only one package to build' and what that package is as the database doesn't currently record the Binaries line from the .dsc.

Revision history for this message
Celso Providelo (cprov) wrote :

promoted to *critical* since the lack of proper P-a-s support is going to be exposed via build-failure-notification.

Changed in launchpad-buildd:
importance: Medium → Critical
status: Unconfirmed → Confirmed
Revision history for this message
Celso Providelo (cprov) wrote :

bug #35283 (native-P-a-s ) task may be required to implement it properly, or we can try to workaround the P-a-s file format and parser for now.

Revision history for this message
Matthew Paul Thomas (mpt) wrote :

<kiko> mpt_, the PAS bug is being solved by me, I guess.

Changed in soyuz:
assignee: nobody → kiko
Revision history for this message
Christian Reis (kiko) wrote :

Landed in RF 4001.

Changed in soyuz:
status: Confirmed → Fix Committed
Celso Providelo (cprov)
Changed in soyuz:
status: Fix Committed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.