Proposed publish_package capabilities

[christophergandrud]

publish_package() should be able to:

• Check that the current build is the most recent checked build (likely based on comparing the Background: package_commit_sha IQSS Report Card element to the most recent commit and if there are any uncommitted changes. If not, then FAIL prompting user to check_best_practice).

• Push to GitHub with a new git tag based on Background:package_version in the Report Card.

• Push to Dataverse with the dataverse package. (need to check if it is possible to create a new version).

• Push to CRAN

Each of these can be disabled based on user publishing needs.

[pdurbin]

@christophergandrud as https://github.com/ropensci/dvn indicates "Many dataverse servers now use Dataverse version 4.0 or greater. A separate package, dataverse, is being developed to work with these newer installations." This is true of https://dataverse.harvard.edu and a good number of other installations listed at http://dataverse.org

At http://dataverse.org/developers we link to a "Client Libraries" page at http://guides.dataverse.org/en/latest/api/client-libraries.html which says, "https://github.com/IQSS/dataverse-client-r is the official R package for Dataverse APIs."

This is a long way of saying that you should use the "dataverse" R package, not the "dvn" package. Thanks! 😄

[christophergandrud]

Thanks for pointing out this error. . . . corrected

[christophergandrud]

I need to dig through this in more detail, but do you know off the top of your head if it's possible to publish an entirely new version of an existing Dataverse dataset from dataverse-client-r?

[pdurbin]

Hmm, not sure. I'm asking at IQSS/dataverse-client-r#13

[leeper]

In terms of R, the publish_dataset() function should let you release a new version of a dataset. I've also implemented the native add file endpoint via add_dataset_file() which would be needed to actually update the file contents themselves (unless you want to use the SWORD API - add_file() + publish_sword_dataset()).

I am still working on the update file endpoint (via a function update_dataset_file()), which you would need for updating a file. Unfortunately R is misbehaving a bit.