Bug#698057: britney: Move/refactor parts into a sep. module
On Sun, 2013-05-05 at 22:13 +0200, Niels Thykier wrote:
> On 2013-01-13 16:38, Niels Thykier wrote:
> > I have written a series of patches to reduce the size of britney.py.
> > In total, these patches removes 274 lines from britney.py[1]. There
> > are some refactoring patches interleaved with this that I thought
> > would be useful to do, but I can take them out if preferred.
> >
> > Reviews greatly appreciated,
Thanks for these, and sorry for not looking over them properly earlier.
On the whole, they look good; thanks. At least one patch would have been
easier to review if a "-w" version had been provided as well. :) A few
queries / comments:
Firstly, I'm assuming that the resulting britney produces the same
result from the testsuite.
+ same_source is an opt to avoid "load global".
Could we avoid using shorthand here? I'm assuming "optimisation" and
something along the lines of "the overhead of forcing a lookup of a
function from the global namespace"; the expansion of the latter is
probably too far in the other direction though. :(
+ This method returns the full dependency tree for the package
+ `pkg`, inside the `arch` architecture for the suite `suite`
+ flatterned as an iterable.
I was going to grumble about the quoting style here, but then realised
it's already in use in other parts of the code; I still don't really
like it though. We appear to have at least three different quoting
styles in use in various parts of the code. :-(
+ If given an iterable it returns a filtered iterator, otherwise it
+ returns a function to generate filtered iterators. The latter is
+ useful if the same filter has to be (re-)used on multiple
+ iterators that are not known on before hand.
"beforehand" is a single word; there's another occurrence in the same
file.
+ The tree (or grapH) is returned as an iterable of (package, arch)
+ tuples and the iterable will contain (`pkg`, `arch`) if it is
+ available on that architecture.
s/grapH/graph/
Regards,
Adam
Reply to: