[Bioc-devel] Need for BiocGenerics?
Kasper Daniel Hansen
kasperdanielhansen at gmail.com
Mon Sep 26 16:54:59 CEST 2011
On Mon, Sep 26, 2011 at 10:45 AM, Vincent Carey
<stvjc at channing.harvard.edu> wrote:
> I didn't mean to imply that the situation can't be improved, but I still
> believe costs and benefits of changes of this sort are hard to reckon. For
> me a real unknown is the added managerial cost of centralization of generic
> definitions. If lots of people have write privilege over the generic
> repository, I can imagine nontrivial confusions; if management is delegated
> to an individual or small group, there can be discouraging bottlenecks. Our
> ad hoc approach has worked but with blemishes. I'd like some evidence that
> a more systematic approach would have fewer blemishes. In theory it would,
> but the practical solution requires some decisions on consensus and process
> that will be costly to make correctly. Is there a language in which the
> problem area has been addressed in a more satisfactory way?
> My gut reaction is that if people really wanted to smooth things out with
> bio-oriented generics, they'd agree with one another that the generic should
> have a specific form, and ask a Biobase maintainer to have it defined in
> Biobase. Could we do some case studies with this approach -- "strand" might
> be a reasonable initial case study -- and then decide on whether an
> additional package will be better?
(1) Having a separate package would make this very lean. I don't
think people consider depending on Biobase a lean solution, but I
could be wrong.
(2) Your point about who and how it gets decided that a generic is
being put in a central package is well taken. I am certainly not
suggesting that all generics should go there automatically, but my
feeling is that this should be somewhat easy to decide, since it
really just requires the community to agree that some method is likely
to be useful across more than one package.
I do agree that it is hard to predict whether the effort will be worth
it. My only 2cp is that it looks like the effort is small.
More information about the Bioc-devel