launchpad-dev team mailing list archive

[Aaron Bentley <aaron@xxxxxxxxxxxxx>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 11-06-21 12:32 PM, Jelmer Vernooij wrote:
> It is possible to build a specific revision if you want to, you can
> specify one in the recipe, like so:
> 
> # bzr-builder format 0.4
> ubuntu:iprint 3
> 
> to build revno 3 of ubuntu:iprint

Yes.  In fact, manifests are recipes and can be used to reproduce a
specific build exactly.

> Still - it will be harder to e.g. write a query that returns all source
> package releases built from the last X revisions in a branch.

Yes, you do get more tables in the query.

>> That seems like a different argument.  Perhaps they are complex, but
>> that has nothing to do with the need for traceability.
>>
>> It would be sub-optimal to have two different ways of determining the
>> built revision.
> I don't see how manifests really provide traceability - the only thing
> they know is what revision from what branch was built.

That is what I mean.  I believe this can be vital information.

> If we keep track
> of that data in another manner anyway, how do manifests help with traceability?

They don't, but it would be sub-optimal to have two different ways of
determining the built revision.

> Most of the interesting traced data is either in the build logs, or in
> the build table anyway (requester of the build, target archive, etc).

The revision id won't be in any of those, IIRC.

>>> Most of the code would be shared.
>> To me, it looks like new code in the buildd, new code in the UI, new
>> database tables.  We use recipes to link everything together, so
>> eliminating recipes means a lot of rewriting.
> The recipe and manifest only complicate matters here - they
> aren't necessary for bzr-builder to be able to build from a branch.
> What benefit would having a recipe have here?

- From a feature development perspective, less work.  From a maintenance
perspective, less code and fewer bugs.  From a user perspective, a
simpler model.

> Hiding recipes at the UI layer for build-from-branch builds is going to
> make things more complicated.

I think having two different features is more complicated.  But I think
that we might need to have unnamed recipes in order for this to work
smoothly.

> Recipes make it
> harder to find back the relevant revision and branch that the build was
> created from.

Harder in the sense of more tables in the query.  Not harder in any
significant sense.

> Manifests don't add extra traceability.

No, but having two different ways of producing source packages makes it
harder to trace the origins of a package.

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAk4A2Z8ACgkQ0F+nu1YWqI1/qgCeJ81DlwR/PFXTQ6O5Q0D/cnSD
N4gAnRoV7xzNuCZtjt2lVBJbt3c0NS5A
=XXs8
-----END PGP SIGNATURE-----