Clean up / fix PEcan.ED2 vignette
infotroph opened this issue · comments
It looks like that very large folder of .h5 files is used in the vignette. I put it in .Rbuildignore for now, but if there isn't already an issue open about fixing up this vignette, you should open one. There's no way CRAN will like all those .h5 files being necessary for a vignette and there are other warnings from
check()about it.
Originally posted by @Aariq in #3135 (comment)
This issue is stale because it has been open 365 days with no activity.
A few more details:
- The folder in question (
models/ed/vignettes/ed_run_data/) is created during vignette execution, not an artifact that we track in Git. - The vignette is currently not run at package build/check time, because some required boilerplate lines are missing from
running_ed_from_R.RmdandDESCRIPTION. Correcting that would be a relatively easy first step on this issue. - Once those are fixed, R CMD check might start complaining that it can't find these files as part of its check whether vignettes can be rebuilt from inst/doc. Or it might not, -- it's possible the vignette will create them there when run.
- If run time / file size are still an issue once the vignette is otherwise working, our options are (1) make examples smaller; (2) precompute the expensive parts; (3) decide they're worth the time/space and ignore the warnings.
I would recommend precomputing the vignettes. E.g. https://ropensci.org/blog/2019/12/08/precompute-vignettes/
And make sure you .Rbuildignore any large files. CRAN does not like large files in inst/