Conversation
amc-corey-cox
left a comment
There was a problem hiding this comment.
This looks good to me. Claude pointed out to me that we now have a few files named in linkml_files.py that we just deleted so they now don't exist. Might be nice to clean that up but it's just a nit. This is ready to go.
| @@ -1,3 +0,0 @@ | |||
| # rdf | |||
There was a problem hiding this comment.
Removing rdf/ and sqlddl/ leaves dangling references in linkml_model/linkml_files.py, which is part of the public API:
Format.NATIVE_RDF,Format.RDF,Format.SQLDDL(enum entries, lines 41/48/52)_Path.NATIVE_RDF,_Path.RDF,_Path.SQLDDL(_PathInfomappings, lines 67/74/78)Format.SQLDDLis in theMETA_ONLYtuple (line 100)ModelFile.model_rdf()andModelFile.rdf()accessors (lines 230/250)
After this merges, calls like LOCAL_PATH_FOR(Source.META, Format.RDF) will silently return paths to files that no longer exist. tests/test_linkml_files.py:52 still references Format.RDF; the test happens to pass only because it asserts on URL-string construction, not file existence.
Could you either drop these enum entries / accessors (with a short deprecation note in CHANGELOG/release notes) or keep the directories? Whichever way, the in-repo API surface should match what's on disk.
There was a problem hiding this comment.
I will wait for @dalito to re-assure me that removal is the right path over "revival" and then I will deal with this! Thank you @amc-corey-cox
|
|
||
| from .types import Integer, String | ||
|
|
||
| metamodel_version = "1.7.0" |
There was a problem hiding this comment.
This and the other generated Python files that I looked at report a metamodel version of 1.7.0. I think this should be updated to 1.11.0.
|
One minor comment above about |
Regenerates the generated artifacts under
linkml_model/against the current source schemas inlinkml_model/model/schema/for the upcoming 1.11.0-rc3 release. Two real changes plus some cleanup:Picks up the
permissivle→permissibletypo fix from Fix mistyping in MatchQuery description #249. That PR fixed the typo inmeta.yamlbut didn't refresh the downstream generated artifacts. This PR propagates the fix intometa.py,jsonld/meta.jsonld,jsonschema/meta.schema.json,protobuf/meta.proto,owl/meta.owl.ttl,shacl/meta.shacl.ttl, andsqlschema/meta.sql.Populates
linkml_model/array.py. The file existed but was empty —array.yamlhas been part of the schema source for a while but the Python module was never generated from it (see Fix equals_string on enum-ranged array_linearization_order #251). Now it ships properly (285 lines).Removes stale
linkml_model/rdf/andlinkml_model/sqlddl/. These directories haven't been refreshed since 2023, are not part of the currentgen_project_config.yamlflow, and were drifting (e.g. still contained the oldpermissivletypo). See linkml/linkml-model#<NEW_ISSUE> for the rationale and a heads-up to downstream consumers. The currentrdf/-equivalent surface is covered byowl/,shacl/, andjsonld/; the SQL output lives insqlschema/. See [LinkML-model] Remove stalelinkml_model/rdf/andlinkml_model/sqlddl/artifacts linkml#3475.