IRC channel logs


back to list of logs

<civodul>i learn Guix has only one package for "atomistic simulation", which makes it less appealing than EasyBuild (24), Spack (19), etc.
<civodul>(for someone into that, that is)
<rekado>I’m slowly adding chemiobioinfo packages
<rekado>the one atomistic simulation package we do have is lammps, isn’t it?
<civodul>i think so, or is it GROMACS?
<civodul>my intern's repo2docker contribution is still pending: :-/
<civodul>for once i feel like i'm on the other side of the patch queue :-)
<civodul>anyway, feel free to add thumbs up to the thing above
<civodul>i was told that Guix should consider applying for NumFOCUS affiliation:
<civodul>"The Reproducible Execution Environment Specification":
<civodul>i knew repo2docker but i didn't know there was a name, a spec, and all
<zimoun>civodul: nice slides!
<zimoun>I think, there is a bias about Repo Accurate package data.
<zimoun>I am not convinced that CRAN 85% can be compared with PyPI 31%. Because of the heterogeneity of inputs, native-inputs, propagated-inputs. Other said, CRAN over-represents “simple” packages and PyPI over-represents complex packages (more depedencies and so on)
<civodul>zimoun: yes, those accuracy measurements are tricky
<civodul>still, we can conclude that CRAN packages usually work out of the box
<civodul>as shown by rekado's auto-importer for R
<civodul>whereas PyPI packages "usually" need more work
<civodul>but as you write, that's also because more of the things on PyPI depend on C/C++ code, for example
<zimoun>Yes, I agree. :-)