Hi Ben,
On Thu, Apr 07, 2016 at 01:52:19AM +0100, Ben Hutchings wrote:
> > For the packages in Debian Science and Debian Med I tend to think that
> > it accommodates a bunch of packages that mostly are of historic value
> > now. People may use them to compare how well their new methods compare
> > against the old stuff
> [...]
>
> Given the low quality and lack of unit tests in many scientific
> applications, how confident can we be that the 'old' packages (that
> have now built with newer toolchains and libraries) actually still
> produce the same results they used to? If we are not, even that
> historic value is lost.
Good point. To address this the Debian Med project is running a GSoC
project[1] and first autopkgtests are written. Moreover I have a UDD
query detecting aging packages[2] of a Blend (= not only maintained by
the Blends team but just in the scope of a Blend). I care for those
packages usually to be uploaded at least every second year (so about
once per release) and I can confirm that every single upload was more
than just a simple rebuild and has uncovered sometimes severe issues
(which also has lead to removal of packages).
So at least as far as it concerns packages in the focus of Debian Med
team I think we do our best to keep also old (in terms of not likely
to be changed) software in a sensible shape.
Kind regards
Andreas.
[1]
https://wiki.debian.org/SummerOfCode2016/Projects#SummerOfCode2016.2FProjects.2FBioToolsTesting.Continuous_Integration_for_all_biological_applications_inside_Debian
[2]
https://anonscm.debian.org/cgit/blends/website.git/tree/misc/sql/aging_packages_of_blend
--
http://fam-tille.de
