While text-based data sources, given their tiny size, fit well in a package and a git repo, they do evolve, and this does not blend in well with evolving functionality of the {n2khab} package. An example is the change brought about by PR https://github.com/inbo/n2khab/pull/182.
Much like the management of the binary data sources, the vc-formatted data sources that are returned in R by {n2khab} will better be managed as a frictionless data package in Zenodo (types, env_pressures, namelist); keeping them together as one Zenodo data source since the (multilingual) namelist is common between types and env_pressures. (When considering tsv-files created with git2rdata::write_vc() in this context, see here and here.)
a straightforward approach to accomplish this and which also continues the version control of these text-based data sources, will be to move the misc/generate_textdata project to a separate GitHub repo, with a webhook to trigger a Zenodo release upon a GitHub release. EDIT: this is questionable; by default a GH release will post the whole repo to Zenodo, not just a resulting frictionless package.
In this way {n2khab} can support multiple versions of these text-based data sources, much in the same way as for other data sources. Furthermore, version interdependency between data sources should then be imposed since a specific version of a (binary) data source works with a specific version range of the text-based data sources, but not all of them.
While text-based data sources, given their tiny size, fit well in a package and a git repo, they do evolve, and this does not blend in well with evolving functionality of the {n2khab} package. An example is the change brought about by PR https://github.com/inbo/n2khab/pull/182.
Much like the management of the binary data sources, the vc-formatted data sources that are returned in R by {n2khab} will better be managed as a frictionless data package in Zenodo (
types,env_pressures,namelist); keeping them together as one Zenodo data source since the (multilingual) namelist is common betweentypesandenv_pressures. (When considering tsv-files created withgit2rdata::write_vc()in this context, see here and here.)misc/generate_textdataproject to a separate GitHub repo, with a webhook to trigger a Zenodo release upon a GitHub release. EDIT: this is questionable; by default a GH release will post the whole repo to Zenodo, not just a resulting frictionless package.In this way {n2khab} can support multiple versions of these text-based data sources, much in the same way as for other data sources. Furthermore, version interdependency between data sources should then be imposed since a specific version of a (binary) data source works with a specific version range of the text-based data sources, but not all of them.