Refactoring model so we could avoid shortcoming of linkml model #257
Conversation
|
@candleindark please review test fails and what needs to be done to make it green |
since Any also represents List or anything else really.
I thought that it might be used somewhere in the meditor to edit dandiset metadata
but it seems that nothing of that type propagated into dandiset metadata so
it should not be effected
=== Do not change lines below ===
{
"chain": [],
"cmd": "git-sedi 'Union\\[Any, List\\[Any\\]\\]' Any",
"exit": 0,
"extra_inputs": [],
"inputs": [],
"outputs": [],
"pwd": "."
}
^^^ Do not change lines above ^^^
…s) for Locus and Allele So far I see no values with such data types used at all in dandi archive so we should not break anything An alternative would be to just make it a List[Identifier] always and not bother allowing for the dance of singular-vs-plural.
That would avoid Union of a list with non-multivalued identifier. And we already have data with string value describing the genotype -- it is not really an identifier!!!
yarikoptic
force-pushed
the
rf-unionany
branch
from
b2205ac to
eecd2f6
Compare
|
I rebased and force - pushed. @candleindark do you think we would still need it? |
|
did anyone check if an updated metamodel was released in linkml as chris mentioned in that original post on linkml. |
The personinfo.yamlid: https://w3id.org/linkml/examples/personinfo
name: personinfo
prefixes:
linkml: https://w3id.org/linkml/
imports:
- linkml:types
default_range: string
classes:
Person:
attributes:
id:
full_name:
aliases:
range: Any
any_of:
- range: string
multivalued: true
- range: integer
phone:
age:
Container:
tree_root: true
attributes:
persons:
multivalued: true
inlined_as_list: true
range: Persondata.yamlpersons:
- id: ORCID:1234
full_name: Clark Kent
age: "32"
phone: 555-555-5555
aliases:
- superman
- man of steel
- id: ORCID:4567
full_name: Lois Lane
age: "33"
aliases: 42Running |
candleindark
left a comment
candleindark
left a comment
There was a problem hiding this comment.
As commented above, this seems to be no longer needed.
|
@yarikoptic I think we can close this one since it is no longer needed as indicated at #257 (comment). |
Shortcoming in question:
AFAIK this should result in no changes necessary to metadata records since none of those "features" was used. So even though we break the model, I think we could just go with "patch" level progress forward.
Individual commits have more information.
TODOs also