Enable full control of data entity metadata

[clnsmth]

Currently, some data entity metadata cannot be defined by users. For example, formatName values are automatically determined within a subroutine of make_eml() and there is no mechanism to override this process. This is obviously an issue when the estimated values are incorrect, or when it's not even reasonable to approximate the values.

A possible solution is to provide an optional entities.txt template for defining entity descriptions, physical metadata, and other information that doesn't fit into the existing set of templates or which would be too onerous to supply as arguments to make_eml(). It's optional in the sense that if missing, make_eml() will continue working as it currently does, but if the template is present, make_eml() will defer to the set values defined therein.

Here's a draft of the entities template for review and comment: https://docs.google.com/spreadsheets/d/1LGuv2gauQXDwrZzYvtbYIWJN5F4uMDGRChf9FyqCbpw/edit?usp=sharing

This template will help resolve #14, #15, #45, #60, #107, #120

Thoughts?

[joefutrelle]

@clnsmth thanks for this. I took a look at the draft template and I have a few questions about it:

• It's in long form rather than wide form. I don't have a strong preference but I'm wondering why you chose that format when the other templates are in wide form.
• Is url required? Not all submitters are going to be able to stage files to public URLs.
• Will make_eml expect all entities to be present in the template in order to apply it, or would it be possible to specify metadata like this for some entities and have the rest be computed? (obviously preferring the latter)

I will follow up with #45 as well.

[clnsmth]

Thanks for these helpful questions and comments @joefutrelle.

• The long form of the template is proposed as a way to handle the variance of properties across EML data types. For example, spatial raster and vector types include the horizCoordSysName property whereas data table and other entity do not.
• No, url is not required. Your question is a good reminder that "required" and "optional" properties will need to be made clear in the documentation.
• make_eml will not expect all entities (which I'm referring to as properties above) to be present in this template. Users will be able to manually specify values when they want, but ignore it otherwise.