-
Notifications
You must be signed in to change notification settings - Fork 2
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
How should we bundle R and Bioconductor? #16
Comments
Running:
shows:
|
From the above we probably need to install OpenBLAS first to be able to use the external-lapack option which is needed to match our build scripts method.
|
I now have three OpenBLAS builds. Each one built using a command like:
|
From reading the Spack docs about their R install method, Spack sets the:
option so the recommended packages that the R developers expect to be installed by default are not. Not sure at the moment how to work with this. This is the current config step for R builds on the clusters:
|
As something to compare against, I've built the base version of R 4.2.2 (the current released version) using our current build-scripts method. This includes the recommended packages so we get the following. before adding our set of extra packages:
Our base Spack build will need to do the same. |
Trying a build using:
|
It didn't like that:
and off is not valid either! |
Try with |
This is working?
|
I've now been able to check what the R binary is linked against:
The OpenBLAS is the serial version created earlier so the build looks like it worked as I wanted. Next step is to see if R actually runs! |
These are the current set of R recommended packages that are expected to be installed with each R installation:
They should all be available in Spack with names of the form r-package so r-codetools for example but all names are lowercase. So to install them I think we just need to do:
|
So running the above has installed MASS in a separate library directory:
I thought it would put a symlink in the main R library directory but it doesn't. This is obviously not the correct approach! |
So there are now 2 module files for R:
and the MASS one adds:
to the beginning of the R_LIBS path. If we did this for all the approx 280 additional R packages we have the R_LIBS path would become massive as would the number of modules to load in the R bundle. This is not what we want. |
We need to decide how we bundle R plus Bioconductor and all the additional R packages we use on the clusters. We need an equivalent of doing:
The text was updated successfully, but these errors were encountered: