Re: A new metric for source package importance in ports
Hi,
Quoting Dmitrijs Ledkovs (2013-11-28 01:15:06)
> On 28 November 2013 00:04, Steven Chamberlain <steven@pyro.eu.org> wrote:
> > I also find it interesting to see openjdk-7 listed but not gcj; or even
> > gcc-4.8. Was this computed for jessie or sid?
>
> I guess implicit relationships are not considered: build-essential
> build-dependencies, and essential dependencies. I would expect for
> packages in those to sets have the highest rank, since,
> hypothetically, all packages in debian build-depend & depend on those.
Steven was looking at the second graph which (in contrast to the first graph)
makes the assumption that essential:yes and build-essential are already
available somehow (for example by having cross compiled them) and thus do not
need to be recompiled to bootstrap the port.
gcj and gcc-4.8 is part of the packages which are drawn in by creating a
co-installation set of essential:yes and build-essential packages. Therefore
they do not appear in the second graph.
Since this co-installation set is an input to the algorithm of creating the
second graph, they implicitly receive the highest rank. For the same reason you
will also see them being assigned the highest rank in the first graph which
does not assume that essential:yes and build-essential do not have to be
recompiled.
Implicit dependency relationships are considered by both algorithms to
calculate the strong dependencies and the dependency closure of source and
binary packages. My code uses dose3 to do the required calculations.
cheers, josch
Reply to: