Provide processed version number

[sdruskat]

There are potentially many "version identifiers" for a single version (timestamp, version string (v1.0.3), etc.). For publication, we need one version identifier.

Includes deduplication, e.g., cff-provided version vs. tag vs. ...

In this processing step, this version identifier will be defined. This processing step should also allow a configuration, where a reserved tag or similar is used to trigger the workflow, which will then, however, extract the real version from the tag. This real version will be used for publication, and could also be used in post-processing to tag the published commit.

[led02]

With #154 almost merged, we can continue here...

• Implement a processing step "set_version"
• configure which alternate version to select

thoght: This could also implemented in an abstract way as a processor (e.g., not only making the precedence configurable but also the key for which this rule should be applied). Also, this could be already decided during the merging of the models.

[process]
use = [ "select_alternative" ]

[[process.select_alternative]]
path = "version"
precedence = [ "cff", "codemeta" ]

[jkelling]

I think the selection of the prioritized version number should be considered part of the curation step, not the processing.

@led02 's idea of allowing users to select harverster precedence on a per-property basis sounds good. This would kind-of an automated pre-curation, or choices made during curation, which can be pushed back into the processing step to provide custom pre-selection on subsequent processing and curation runs. I think there should not be specific code for the version number.

[poikilotherm]

Reopening because we still need to discuss how to do this. (Closed by accident)

Also, we need to take into account version numbers from the context/environment (CI provides some of this data when using tags), git harvester and (potentially) project files (Maven POM, pyproject.toml, etc)

[jkelling]

Reopening because we still need to discuss how to do this. (Closed by accident)

I think we should move the this issue to the alpha milestone then, because the basics are already there. The discussion will be about more advanced configuration options to automate part of the curation step.

Also, we need to take into account version numbers from the context/environment (CI provides some of this data when using tags), git harvester and (potentially) project files (Maven POM, pyproject.toml, etc)

Taking into account other sources does not look like an open question to me: the way to add support for each thing you listed is by adding a corresponding harvester. Automatic selection from among conflicting data sources is part of curation-automation question.

[sdruskat]

Moved to Alpha milestone.