Bug#1054657: transition: r-bioc-biocgenerics

[Charles Plessy <plessy@debian.org>]

Le Fri, Nov 10, 2023 at 11:30:13PM +0100, Sebastian Ramacher a écrit :
> 
> Can you please clarify whether you are talking about new dependencies or
> reverse dependencies above? Thanks.

Hi Sebastian,

I am talking about the reverse-dependencies of the packages that we need
to upload to NEW.  These are new dependencies of packages that need to
go through the Bioconductor transition.

Here is an example:

In Bioconductor 3.17, the new package S4Arrays was depended on by
DelayedArray.  This reverse dependency of a new package in Bioconductor
is a new dependency on a new package in Debian (r-bioc-delayed array
started to depend on r-bioc-s4arrays).  The upload of r-bioc-s4arrays
during the 3.17 transition is one of the causes of the delays.

Last week I thought that looking at the reverse-dependencies of packages
that are new in Bioconductor 3.18 would be a good heuristic to find if
we will need to introduce new packages in Debian.  This is the main
reason I expresesd my thoughts in terms of reverse-dependency of new
packages, instead of new dependencies on new packages.  However
yesterday me and Andreas figured out that this was not a good heuristic.

Here is an example:

In Bioconductor 3.18, DelayedArray starts to depend on SparseArray.
However, SparseArray is not new in Bioconductor 3.18.  It was
introudced in Bioconductor 3.17 but we did not notice and package it
because nothing was depending on it.

In conclusion, taking the point of view of reverse-dependencies of new
packages in Bioconductor was not as fruitful as I thought and I will try
to stick to the point of view of new dependencies on new packages in
Debian to avoid further confusions.

I hope it clarifies.

-- 
Charles