Versioning scheme for terms and entire standard about ets

[fdschneider]

Thanks Caterina! I tried to implement these comments in the Standard v0.7.1.
However,

• issues with versioning are not really resolved.

I think the dateIssued and version were left empty where the term forwards 1:1 to DwC. However, this might not be best practice, since it might be important to keep a dateIssued within our data standard if the original definition is altered on the side of DwC. I did not fix errors or unclear definitions in those, because it would break the dependency (that also includes the semicolon use and heterogeneities in definitions and comments).

The version column is supposed to give a version for each term, but we have not defined how the termed versioning relates to the versioning of the entire data standard. Therefore, I would remove this from the html view and add it as an open issue. Probably we should implement a kind of semantic versioning on the term level (0.0.1 for typos and insignificant changes, 0.1.0 for meaningful changes; 1.0.0 for incompatible changes) and issue a new vocabulary version increment 0.1 each time one or several meaningful changes are issued, and increment 1.0 each time one or several major changes are issued.

This also concerns the use of

• 'dateModified': This is supposed to report the last activity on this term, but has not consistently been recorded.

• 'Replaces' and 'ReplacedBy' : if a term has been altered beyond compatibility, the old/new version might be referred here using the versioned path of the term.

• 'Deprecated' : if an old version of a file is kept in the standard, but is not supposed to be used ( should be combined with 'ReplacedBy')

• 'DateIssued' should be the date when the term first was defined, and thus remains rather static.

• We need to define a date format. This is messed up.

• The valueType field aims at the use in R, which is why I stuck with the value types provided there.

• I also removed the entries 'relation' and 'relationSource'. Those were introduced with the idea of defining an own trait thesaurus with this vocabulary (i.e. issueing an own identifier), but now we sell these terms to be used in a metadata definition of traits, pointing to external thesauri or ontologies.

• the semicolon problem: cant be solved thanks to Excel. You have to switch your Windows number format to 'English (United States)', or 'en-US' to open and save the file. We might need to transfer this to an XML-based ontology scripting language at some point to avoid these problems (see #1). For now I suggest that we handle the csv file very carefully!

[fdschneider]

I did

• remove the term-level versioning from the source file.
• add a 'FirstIssuedIn' field, which points to the version of the standard where this term has been added. The 'Deprecated' and 'Replaces'/'ReplacedBy' fields are sufficient for versioning.
• add a statement on versioning in the 'about' and 'contributing' page