Change bulk upload (via wizard or API) to allow for traits not in the trait_covariate_associations table

[gsrohde]

See terraref/computing-pipeline#193 (comment) for discussion of how this should work.

[gsrohde]

Changes for the API have been implemented (but not deployed). This changes the API behavior as follows.

[From issue #193:]

1. Any column heading that matches an existing variable name will be recognized as a trait variable except (a) if it is listed in the trait_covariate_associations table as a covariate; (b) in the unlikely event that it also happens to match one of the canonical column headings—citation_author, citation_year, citation_title, citation_doi, cultivar, species, site, treatment, utc_datetime, local_datetime, date, yield, n, SE, notes, access_level, or entity. (This rule is mainly to prevent everything from breaking if someone somewhat perversely adds one of these unlikely names as the name of a variable.)
2. The use cases for adding a row to the trait_covariate_associations table are (a) we wish to recognize a column heading as the name of a covariate variable associated with some (or some set of) trait variable(s); (b) we wish to enforce a requirement for certain trait variables that certain associated covariates always be specified. (For example, on https://betydb.org, leafT must always be specified when trait data for Vcmax, quantum_yield, Amax or certain other variables is uploaded.)

In addition, it is an error if there is a variable name listed in the heading that corresponds to a covariate in the trait_covariate_associations table without there also being a corresponding trait name.

[gsrohde]

Fixed by release 4.15.

[gsrohde]

Re-opening. This feature is not yet working properly.

[gsrohde]

@dlebauer I made a note that this was not yet working properly, but I don't remember who reported this or any other details.

[dlebauer]

If I have a csv upload with a column name that matches a name in the variables table, it should insert the values into the traits table. The absence of a covariate is irrelevant. The traits covariates associations table is useful when there is a covariate. But it should not prevent inserting variables that do not require a covariate.

…

On Fri, Jun 2, 2017 at 5:34 PM Scott Rohde ***@***.***> wrote: Assigned #459 <#459> to @dlebauer <https://github.com/dlebauer>. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#459 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAcX55N3sIJNad7HFeTt7DmIcNXsiNCIks5sAIVrgaJpZM4Ku4--> .