ashb commented on code in PR #72043:
URL: https://github.com/apache/airflow/pull/72043#discussion_r3956545480


##########
go-sdk/adr/0007-native-dag-interface.md:
##########
@@ -0,0 +1,140 @@
+<!--
+ Licensed to the Apache Software Foundation (ASF) under one
+ or more contributor license agreements.  See the NOTICE file
+ distributed with this work for additional information
+ regarding copyright ownership.  The ASF licenses this file
+ to you under the Apache License, Version 2.0 (the
+ "License"); you may not use this file except in compliance
+ with the License.  You may obtain a copy of the License at
+
+   http://www.apache.org/licenses/LICENSE-2.0
+
+ Unless required by applicable law or agreed to in writing,
+ software distributed under the License is distributed on an
+ "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ KIND, either express or implied.  See the License for the
+ specific language governing permissions and limitations
+ under the License.
+ -->
+
+# 7. Native Dag interface
+
+Date: 2026-09-07
+
+## Status
+
+Proposed. Supersedes the interfaces proposed in #67155 and #70158, reshaped by 
the design review
+on #72043. Nothing below exists on `main`.
+
+## Why
+
+A native Dag is authored entirely in Go: the author owns the schedule, the 
tasks, and the
+dependencies, and the SDK serializes all of it into the Dag JSON a Python Dag 
would produce. There is
+one Dag type, as in Python — the Mixed Lang case registers task handlers 
instead
+([ADR 6](0006-mixed-lang-task-handler-interface.md)), because it defines no 
Dag. Dependencies between
+Go functions have to be typed rather than looked up by task ID, and a Dag 
should read like Go rather
+than like transliterated Python.
+
+## Example
+
+Both examples build the same graph. Which form an author writes depends on 
whether the edge carries a
+value.
+
+### Data dependencies: the TaskFlow equivalent
+
+`airflow.Inputs` passes an upstream's return value in and declares the edge in 
one call, the way
+calling one TaskFlow function with another's output does in Python.
+
+```go
+dag := airflow.Dag(airflow.DagSpec{DagId: "etl", Schedule: "@daily"})
+
+extracted := dag.Task(extract)
+transformed := dag.Task(transform, airflow.Inputs(extracted))
+dag.Task(load, airflow.Inputs(transformed), airflow.TaskSpec{Retries: 2})
+
+registry.AddDags(dag)
+```
+
+```python
+# the Python Dag this mirrors
+extracted = extract()
+transformed = transform(extracted)
+load(transformed)
+```
+
+```go
+func extract(ctx context.Context) (Result, error) {
+    return Result{Message: "native Dag data"}, nil
+}
+
+func transform(ctx context.Context, extracted Result) (Result, error) {
+    return Result{Message: "transformed " + extracted.Message}, nil
+}
+
+func load(ctx context.Context, transformed Result) error { return nil }
+```
+
+### Order-only dependencies: the `>>` and `<<` equivalent
+
+`Before` and `After` draw an edge and pass nothing, for tasks that must be 
ordered but exchange no

Review Comment:
   Elsewhere in this PR you have `task.Then(task2)` -- we likely shouldn't have 
both, right? Lets be consistent.



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to