Should we have run_exports on LinkingTo packages?

[mfansler]

_{Note: This isn't specifically an Rcpp issue, but Rcpp is a prominent representative of the class of packages to which this pertains.}

Browsing issues on Bioconda, I noticed that there is some effort to more broadly add run_exports to their recipes, including for Bioconductor recipes. While I'm not so sold on putting it on everything, I am wondering:

Should we be adding run_exports to packages that get used in LinkingTo sections?

Personally, I've never run across an issue reporting having hit an Rcpp runtime incompatibility, but who knows if a user would know where to report it if encountered.

Any thoughts, @conda-forge/r?

[cbrueffer]

Indeed, I don't think I've seen issues either, but who knows. I don't have strong opinions on this either way, but would lean to keeping things as is for now. In Bioconda we have a lot lot more varied ecosystem where incompatibility issues are probably more likely to crop up.

[mfansler]

I just hit one of these with r-lme4 linking against r-matrix (conda-forge/r-matrix-feedstock#42). However, still doesn't seem worth using run exports.