From 691d3c33134a1932d9957c2acaea8685872c2593 Mon Sep 17 00:00:00 2001
From: Michael Hanke <michael.hanke@gmail.com>
Date: Thu, 4 Jan 2024 08:38:11 +0100
Subject: [PATCH] New `git` ontology component with new `GitTracked` mixin

This aims to be the foundation for data models of Git-tracked content.
The availability of a globally unique identifier in this context
allows for great flexibility in model specification. At the same time,
introducing a `identifier:true` property in a class hierarchy can be
problematic, because turning an existing base-class property into an
identifier in a derived class does not seem to be possible
(https://github.com/psychoinformatics-de/datalad-concepts/issues/30).

This approach uses a dedicated mixin class for this purpose.

The identifier itself uses the `meta_id` slot, and is intentionally
distinct from the `gitsha` property. The identifier is composed from
the `gitsha` by adding a corresponding namespace prefix. This keeps
the identifier property uniform across unrelated classes, while
also preserving the semantics of the identifier type.
---
 src/linkml/git-provenance-schema.yaml | 30 ++------------------
 src/linkml/git.yaml                   | 41 +++++++++++++++++++++++++++
 2 files changed, 43 insertions(+), 28 deletions(-)
 create mode 100644 src/linkml/git.yaml

diff --git a/src/linkml/git-provenance-schema.yaml b/src/linkml/git-provenance-schema.yaml
index 625abbe..5c585d4 100644
--- a/src/linkml/git-provenance-schema.yaml
+++ b/src/linkml/git-provenance-schema.yaml
@@ -14,6 +14,7 @@ prefixes:
   linkml: https://w3id.org/linkml/
   prov: http://www.w3.org/ns/prov#
 imports:
+  - git
   - labeling
   - meta_utils
   - provenance
@@ -21,13 +22,6 @@ imports:
 default_prefix: dlco
 
 
-slots:
-  gitsha:
-    description: >-
-      SHA1 identifier of a Git object.
-    range: SHA1
-
-
 classes:
   # TODO orthogonalize/consolidate with datalad-dataset:Commit
   # the relevant class is not this one, but the
@@ -37,31 +31,11 @@ classes:
       A revision or an individual change in a Git repository.
     mixins:
       - CommitProvenance
-    slots:
-      - gitsha
-      - meta_id
+      - GitTracked
     slot_usage:
       gitsha:
         description: >-
           SHA1 identifier of the commit object.
-      meta_id:
-        pattern: "^gitsha:[0-9a-f]{40}$"
-        description: >-
-          SHA1 identifier written as a CURIE in an explicit
-          `gitsha:` namespace.
-
-    unique_keys:
-      gitsha:
-        description: >-
-          A commit's GITSHA is a globally unique, content-based identifier
-          of that commit.
-        notes:
-          - The `gitsha` slot could have been labled to be the one identifier
-            of a Commit-type object. However, we use `meta_type` for
-            consistency and use its CURIE-format to assign an explicit
-            namespace.
-        unique_key_slots:
-          - gitsha
 
   CommitProvenance:
     mixin: true
diff --git a/src/linkml/git.yaml b/src/linkml/git.yaml
new file mode 100644
index 0000000..ae568ab
--- /dev/null
+++ b/src/linkml/git.yaml
@@ -0,0 +1,41 @@
+id: https://concepts.datalad.org/git
+name: git
+title: Git concepts
+description: >
+  TODO
+prefixes:
+  gitsha: https://concepts.datalad.org/namespace/gitsha/
+  dlco: https://concepts.datalad.org/ontology/
+imports:
+  - meta_utils
+  - types
+default_prefix: dlco
+
+
+slots:
+  gitsha:
+    description: >-
+      SHA1 identifier of a Git object.
+    range: SHA1
+
+
+classes:
+  GitTracked:
+    mixin: true
+    description: >-
+      Mixin class for any ressource tracked by Git, thereby having a unique
+      `gitsha` identifier.
+    slots:
+      - gitsha
+      - meta_id
+    slot_usage:
+      meta_id:
+        description: >-
+          SHA1 based identifier in the form of a CURIE with an explicit
+          `gitsha:` namespace prefix.
+        equals_expression: "gitsha:{gitsha}"
+        pattern: "^gitsha:[0-9a-f]{40}$"
+    comments:
+      - Because there can only be a single main `identifier`, this mixin
+        cannot be used on classes that already have or inherited another
+        property declared as `identifier:true`.