dsc branch has odd structure

(IOW, the initial commit is unnecessary)

Right now the importer/ branch manipulators do not need to create the branches, they can assume that they exist. To pre-create the branches we need a commit. They happen to be the same commit because the treeish is the same and the authorship is the we. Perhaps there is some latent bug in the code and the date is wrong, so the commit hashes are fully identical. It's certainly not strictly necessary.

Also the dsc branch is unused by anything. It could be used by build as we
have a git cache of all dsc imported without launchpad, just like the
pristine-tar branches.

On Nov 22, 2017 21:20, "Robie Basak" <email address hidden> wrote:

> Public bug reported:
>
> I see an "Initial DSC branch." empty commit shared between both 'debian'
> and 'ubuntu' dsc branches. Then one commit per DSC.
>
> This isn't really a problem, but isn't clean either. Is it possible to
> have the first DSC import just start with an orphan commit in both
> cases?
>
> I mention this now because this is hash-abi-break and won't be able to
> be changed later.
>
> ** Affects: usd-importer
> Importance: Undecided
> Status: New
>
>
> ** Tags: hash-abi-break
>
> --
> You received this bug notification because you are a member of Ubuntu
> Server Dev import team, which is subscribed to usd-importer.
> https://bugs.launchpad.net/bugs/1733895
>
> Title:
> dsc branch has odd structure
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/usd-importer/+bug/1733895/+subscriptions
>

I just hit this while writing tests for import_unapplied_spi(). We have a dependency on the dsc branch existing already, which is (slightly) awkward.

Any objection if I change the code to create the dsc branch on first use if it doesn't already exist? I think this would be cleaner, but also I will eliminate the need for pre-unit-test prep then.

On 22.12.2017 [14:55:01 -0000], Robie Basak wrote:
> I just hit this while writing tests for import_unapplied_spi(). We have
> a dependency on the dsc branch existing already, which is (slightly)
> awkward.

Ah yes, I can see that.

> Any objection if I change the code to create the dsc branch on first
> use if it doesn't already exist? I think this would be cleaner, but
> also I will eliminate the need for pre-unit-test prep then.

Sure, that's fine with me. Would it make sense to just have this be part
of the constructor? It's sort of definitionally part of our repository
structure now.

-Nish

--
Nishanth Aravamudan
Ubuntu Server
Canonical Ltd