Bug#678359: autopkgtest: Depends: @ only works when building the source

[ijackson@chiark.greenend.org.uk (Ian Jackson)]

Martin Pitt writes ("Re: Bug#678359: autopkgtest: Depends: @ only works when building the source"):
> Ian Jackson [2012-06-28  1:01 +0100]:
> > Thanks for working on this.  Looking at it I'm not entirely clear that
> > the patch is right.  I'm quite tired right now and could be wrong but
> > it seems to me that we should be using the tree from the tree argument
> > to read_control.  This should probably become a new constructor
> > argument to Test (and correspondingly doesn't need to be an argument
> > to run_tests or indeed to Test.run).
> 
> Where "This" == the list of binaries that are being produced by this
> source (as a new ctor argument)?

No, the tree.

> > And of course you should use tree.read() not tree.path[1].
> 
> Ah, thanks! Cf. That was not obvious to me.

NP.

> > (And, as a matter of style, Python `portability' stuff notwithstanding
> > I think the code is clearer if you avoid os.path.join and just say `/'
> > when you mean it.)
> 
> It's an old habit of mine, but I'm happy to use / directly.

Heh :-).

> > Perhaps it would be better to generalise the first half of
> > read_control until it can be reused (as a non-class function called
> > something like read_stanzas, I guess).
> 
> Will do that.

Thanks.

Ian.