Circle errors

[cboettig]

Looks like the current circle error is due to to the fact that littler loads so few packages. You could add methods, stats, graphics, and grDevices to the library calls in the niche-modelling, but I find it easier to omit these and run with:

R -e 'rmarkdown::render("manuscript.Rmd")'

That way you get all the packages a usual R session has (unlike r or RScript).

It also looks like you will then need to install the most recent wicket version from GitHub to resolve another error.

That should get you up to the issue at #8.

HTH

[cboettig]

p.s. since rocker/ropensci is such a big image, I'd consider switching to rocker/geospatial:3.4.1 (using the vesion tag for reproducibility) , and add a DESCRIPTION file to this repo like so:

Package: compendium
Version: 0.1.0
Depends: rmarkdown, tidyverse, rgbif, dismo
Remotes: ropensci/wicket@99d3b14

Then I'd have circle-ci run:

run -ti -v $(pwd):/data -w /data rocker/geospatial:3.4.1 R -e 'devtools::install(); rmarkdown::render("manuscript.Rmd")'

[sckott]

made the changes, still getting a wicket error though, not sure why

[sckott]

got a successful run https://circleci.com/gh/sckott/gbifms/58

now trying to get deploy working

[sckott]

working now except deploy is not working, meh, im over it, can't get it to work, can just manually update pdf in repo i guess

[cboettig]

yeah, I usually do that anyway so I don't like having to copy my credentials around. (I do wish GitHub would issue repo-specific PATs instead of global PATs!) Good that it builds on circle, that's always nice for reproducibility!

Anyway, I think this is good to go as a preprint now, yes?

[sckott]

i think so, okay if i submit?

(for deploy, i'll see if can get a deploy to S3 at least, so we have a stable URL to link to)

[cboettig]

yup, submit away! 🚀

For stable URL, let's just activate the Zenodo toggle and tag a preprint release. Of course that gives us only a DOI for the repo as a whole, but anyone just wanting the pdf file should probably use the DOI that goes to the preprint anyway.

[sckott]

Right, makes sense on Zenodo front, will do

[sckott]

thoughts @cboettig ?

[sckott]

computer sci i guess?

[cboettig]

I would go with Life Sci, since I think that's more the 'audience' of the piece.