FS#68577 - [r] prepackaed library is out of date
Attached to Project:
Arch Linux
Opened by M. Greyson Christoforo (greyltc) - Monday, 09 November 2020, 17:58 GMT
Last edited by Antonio Rojas (arojas) - Monday, 09 November 2020, 20:28 GMT
Opened by M. Greyson Christoforo (greyltc) - Monday, 09 November 2020, 17:58 GMT
Last edited by Antonio Rojas (arojas) - Monday, 09 November 2020, 20:28 GMT
|
Details
This package seems to bring with it a number of libraries.
I'm not sure how this list is chosen, but I assume there's
some good reason for these specific libraries to be built
into the r package.
``` $ ls -1 /usr/lib/R/library base boot class cluster codetools compiler datasets foreign graphics grDevices grid KernSmooth lattice MASS Matrix methods mgcv nlme nnet parallel rpart spatial splines stats stats4 survival tcltk tools translations utils ``` These libraries have their own versions which are not captured in the R package version and this can cause problems when they become out of date and then impossible to update because they're rolled into the r package that's not really totally out of date. For example, today the version of this package is 4.0.3 and that's great because 4.0.3 is the current stable released version of R. What's not great is that the stable release version of the KernSmooth R library is 2.23-18 (see https://cran.r-project.org/web/packages/KernSmooth/index.html ) but the KernSmooth built into this package is 2.23-17: ``` $ cat /usr/lib/R/library/KernSmooth/DESCRIPTION Package: KernSmooth Priority: recommended Version: 2.23-17 Date: 2020-04-26 ... ``` This can cause some real problems, like today the latest version of Bioconductor ( http://www.bioconductor.org/ ) can not be installed because it needs the very latest version of KernSmooth, 2.23-18 So how can KernSmooth be updated individually if its bundled into the core r package? Do all these libraries really need to be bundled togehter here? Should they become dependencies that can then be updated seperately so that their versions can be properly tracked? Is the right thing to do just flag the r package as out of date when any of these libraries become out of date and then rebuild it, bumping the release int? |
This task depends upon
Comment by Antonio Rojas (arojas) -
Monday, 09 November 2020, 19:23 GMT
Comment by
M. Greyson Christoforo (greyltc) -
Monday, 09 November 2020, 19:51 GMT
Those libraries are included in the upstream tarball. If you need
a more recent version, you can override them with a locally
installed version.
Ok. Thanks!