[RFC] Inventory refactoring (brisbane)

[Andrew Bennetts]

Robert Collins wrote:
> On Mon, 2009-04-06 at 18:03 +1000, Ian Clatworthy wrote:
> > As this is arguably the critical path for merging brisbane-core
> > and I'm offline tomorrow for medical reasons, can someone else
> > please consider this problem, and if they agree with me, write
> > these missing tests ASAP?
[...]
> 
> My inclination is to look at this as we polish the new format; for the
> following reasons:
>  - the interface tests are not failing without those methods
>  - some blackbox tests may need them
>  - until CHK based repositories are widely available we run a low risk
>    of a method on Inventory not being present on CHKInventory
>  - inventories are not the recommended public interface for tree's 
>    anyway. [they are there and supported but not the defacto interface]
>  - when I checked loggerhead, one of the plugins most likely to need
>    all the inventory data, it worked ok.

I agree with Rob.

There's no risk of regression to non-CHK users here.  And judging from the
passing test suite and actual use of the brisbase-core branch there's
unlikely to be many bugs (if any) for CHK users here either.  Certainly when
landed in bzr.dev it won't be any less functional than it is in
brisbane-core, so I see no reason to worry about this now.  i.e. go ahead
and land.

-Andrew.