[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Uploaders orphaning packages / Conda (Was: Bug#921382: kineticstools: autopkgtest needs update for new version of h5py)


On Tue, Feb 05, 2019 at 02:56:36AM -0500, Afif Elghraoui wrote:
> > What means "interest"?  If it is I should not spend time on packages I'm
> > not using for my own work I can leave Debian Med since I'm not using a
> > single one.  As a Debian maintainer I want to provide a useful system
> > for my colleagues and the motivation to maintain also packages that are
> > not used here is to possibly attract even more people to join the
> > effort.  This attraction of people has worked to some extend (way better
> > than in other teams but admittedly I was hoping for more active
> > contributors).
> Ok, that's another form of interest. My original interest was personal,
> with the willingness to share work I was going to be doing for myself
> anyway. Yours is a little bit different.

Yes.  I'd call this "meta-interest". ;-) 

> > I do not see Conda as a pure
> > competitor to our efforts since there are also different areas where the
> > different packaging efforts are differently suited.
> I think it is a pure competitor for the bioinformatics packages; at
> least the ones I worked on.

I think lots of differently versioned applications in lots of user home
directories will lead to confusion sooner or later.  I have no idea how
Conda works on big cluster installations etc.  But that are details we
might discuss at the sprint.  The fact that Conda exists and is
successful should be a good reason for us do spent some time into it.
> >  However, by all
> > means we should force the effort to package the Conda tools themselves
> > in any case.
> I gave conda as an example.

Sure, but we should package the tools anyway.  If we are packaging
software for biologists and other scientists and they need Conda for
their work we should probably provide it as a tool.

> It has problems of its own, like unbearable
> sluggishness [1,2], but the general idea of cross-platform, unprivileged
> package management is what appeals to scientists.

Fully understand this point - but our admins are suspicious about it.

Kind regards



Reply to: