Re: R 4.0.0 is here, transition is stalled
Hi Charles,
On 26 April 2020 at 17:08, Charles Plessy wrote:
| Le Fri, Apr 24, 2020 at 06:35:58PM -0500, Dirk Eddelbuettel a écrit :
| >
| > But the transition seems stalled. I am a little worried because there is one
| > (for GNU GSL) which has been in "real soon now mode" for half a year.
|
| Hi Dirk and everybody,
|
| I had a look at the collisions listed on the r-api-4.0 transition page:
|
| * auto-gsl through r-bioc-dirichletmultinomial, r-cran-gsl, r-cran-rcppgsl
|
| As you are the maintainer, maybe you can instruct the Release team which
| you think should go first ?
GSL, obviously. Version 2.6 was packaged seven months ago:
https://salsa.debian.org/edd/gsl/-/commit/c45f77e1efe6b567cf1dff9815ab085251d0612f
| * auto-hdf5 through r-bioc-rhdf5, r-bioc-rhdf5lib, r-cran-hdf5
|
| A hdf5 transition was just completed and the current one in the tracker
| is about the next one. Hopefully it means we go first.
|
| * auto-icu through r-cran-stringi
|
| I also do not see mention of this one in bugs.debian.org or in
| debian-release@l.d.o; let's cross the fingers we can go first.
Ok, let's keep an eye on these.
| * auto-netcdf through r-cran-ncdf4, r-cran-rnetcdf
|
| The transition is ongoing, and only two package are holding it: vtk7 and
| qgis, for build issues on mips64el and mipsel. Nevertheless, I see that
| the buildds are running now, so let's cross fingers.
Ok, good news.
| * auto-nodejs through r-cran-v8
|
| Not started, and no visible dicussions…
|
| * python3.7-rm through r-cran-fastcluster
|
| The transition is ongoing and one of the packages holding it is
| python-biopython, because of a build failure on mipsel which is caused
| by the clustalo tool (#956324). Members of the Debian Med team,
| if we want to accelerate the R transition, we need to move the clustalo
| bug out of the way :)
:)
| > Would it be polite to nudge, or not? What is the best process?
|
| Yes, I think that it would be worth contacting the Release team about
| the auto-gsl transition. Would you like to hold it before the R one is
| completed ?
Do be clear, by "hold" you mean have it occur later? No, I think GSL should
go first. (But that is a weak preference. They both need to happen sooner
rather than later and strict order is irrelevant as long as both happen.)
In the meantime I may switch more of my use of R via Debian testing (which is
mostly in CI tests, frequently via Docker) to using the updated 4.0.0 base
container in our rocker/r-base:4.0.0 and install packages directly from CRAN
as needed as I have just done for Rcpp:
https://github.com/RcppCore/Rcpp/commit/1812cdad8c8779aeec8c4d5778f1a37ee89587d8
Thanks for the very helpful emailing and keeping an eye on the transitions.
As for ppc64el, I have not yet heard back from Brazil about the VM account.
Dirk
--
http://dirk.eddelbuettel.com | @eddelbuettel | edd@debian.org
Reply to: