Closed John-Polo closed 1 year ago
Instead of using a skeleton
, I found that if I used remotes::install_github
, I can get my environment created. This may still be a problem for someone else down the line, but I will close this, since I don't need it any more.
Checklist
What happened?
I am trying to build an R package for conda-forge that is currently only on GitHub. I'm new to this process and have little experience with Python or conda. I mostly relied on this post from StackOverflow and some other posts from web search.
The package I want to post to conda-forge is this: https://github.com/ncsu-landscape-dynamics/rpops
I created a YAML with the rpops dependencies and then used:
to create the environment. I then started R and used
to see if R in the created environment would install the package. Installation worked without a problem. I quit R. Then in the lbpops environment ran
The skeleton call works. This is the result:
Instead of the error, I was expecting a package to be created with conda build that would be suitable for uploading to conda-forge.
Conda Info
Conda Config
Conda list
Additional Context
I tried setting the pop.yaml to different versions of r-base and r-landscapemetrics. In this last attempt, I left out r-base, the environment seems to know how to get that and I did not assign a version to r-landscapemetrics, so both of those should be whatever the environment found for resolving the build. I'm not sure I'm interpreting the error correctly, but I think it's telling me that conda build can't resolve the r-base and r-landscapemetrics versions, but those were set by the environment and they came from conda-forge, so I don't know why it can't resolve those packages. I tried old versions of R, like 4.1.3, and also setting r-landscapemetrics < 1.5.6, which is the version on conda-forge currently.