🚸 Broaden name deduplication in `HasType` to search all contexts if no `type` is passed by ishitajain9717 · Pull Request #3850 · laminlabs/lamindb

ishitajain9717 · 2026-08-10T20:40:08Z

Add fallback to search typed objects when root-level deduplication finds nothing. If no type is passed, deduplication first searches root-level objects as before. If nothing is found there, it falls back to searching across all types — preventing silent duplicate creation when a same-named objects exists anywhere in the hierarchy.

Also adds a warning when type=None is explicitly passed for non-type objects, since root-level creation should be intentional and rare.

falexwolf · 2026-08-11T01:52:33Z

Great!

Also adds a warning when type=None is explicitly passed for non-type objects, since root-level creation should be intentional and rare.

Can you condition this on is_type=False? If is_type=True, then this is likely intentional.

ishitajain9717 · 2026-08-11T06:32:26Z

Hi,
It is already conditioned on is_type=False.

…ssed

… fallback

codecov · 2026-08-12T15:00:32Z

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 86.98%. Comparing base (112244e) to head (9033d3a).
⚠️ Report is 9 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3850      +/-   ##
==========================================
- Coverage   91.55%   86.98%   -4.58%     
==========================================
  Files          88       88              
  Lines       15747    15845      +98     
==========================================
- Hits        14417    13782     -635     
- Misses       1330     2063     +733

Files with missing lines	Coverage Δ
lamindb/models/schema.py	`93.91% <ø> (ø)`
lamindb/models/sqlrecord.py	`92.16% <100.00%> (+0.08%)`	⬆️

... and 16 files with indirect coverage changes

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

github-actions · 2026-08-12T15:03:27Z

Deployment URL: https://7dfd74f7.lamindb.pages.dev

falexwolf · 2026-08-13T13:52:38Z


 def suggest_records_with_similar_names(
-    record: SQLRecord, name_field: str, kwargs
+    record: SQLRecord, name_field: str, kwargs, type_val=UNSET


Why do we need to introduce type_val? The kwargs have type in them and you can detect whether it matches UNSET. Why is that not possible?

falexwolf · 2026-08-13T13:55:52Z

    # the below needs to be .first() because there might be multiple records with the same
    # name field in case the record is versioned (e.g. for Transform key)
    if isinstance(record, HasType):
-        if kwargs.get("type", None) is None:


I would have expected something like this:

type = kwargs.get("type", UNSET) if type is UNSET: subset = record.__class__.filter() elif type is None: subset = record.__class__.filter(type__isnull=True) else: subset = record.__class__.filter(type=kwargs["type"])

There is a question whether we should use kwargs["type"] instead of kwargs.get() because we might now actually have a contract that guarantees the presence of type since it would never be popped? 🤔

falexwolf · 2026-08-14T09:27:44Z

                from .collection import Collection
                from .transform import Transform

+                # ensure "type" is always present in kwargs for HasType models so that


Do you understand why this is necessary? My assumption is that every HasType model populates type=UNSET or whatever the user passes. How can it be that there are cases where you need this defensive line?

If this is AI generated: Can you remove it and see if the tests pass?

I think the injection is needed for Project and Reference. Unlike Record, ULabel, Feature, and Schema, which explicitly pop and re-inject type (defaulting to UNSET) in their own constructors, Project and Reference have init methods that pass args, kwargs straight through to super(). So if a user calls Project(name="foo") without type, the key is simply absent from kwargs when BaseSQLRecord receives it.
Please suggest if it makes sense

It's great that you found this!

Project and Reference have init methods that pass args, kwargs straight through to super().

Hmmm. If that's what's happening then we should see type: UNSET in the kwargs, dict, right? I think your reasoning might be incorrect here.

You're right that we should always see type: UNSET in kwargs — that's the contract. But the problem is that UNSET is an internal sentinel. So "type" only ends up in kwargs if the model's init explicitly injects it. Do you agree?

Maybe I missed something in the previous PR: I understood that we changed type = None to type = UNSET in the previous PR. This isn't an internal but a user-facing change. If it wasn't user-facing, how would a user learn about the difference of passing type = None versus not passing type?

I'm convinced that every HasType registry needs to expose type = UNSET to the user also in the docs; and of course, if the user doesn't pass type that needs to get passed into the downstream calling cascade. That should IMO be automatic since UNSET is the default value of the constructor.

falexwolf · 2026-08-14T09:31:29Z

                    kwargs["created_on"] = kwargs["branch"]
            if skip_validation:
+                # strip UNSET just before Django sees kwargs — FK descriptors reject non-model values
+                if isinstance(self, HasType) and kwargs.get("type", UNSET) is UNSET:


The logic here is incorrect: if "type" is not in kwargs the .get() statement returns UNSET and one enters kwargs.pop() even though there is no key "type"!

IMO the same rationale that would make https://github.com/laminlabs/lamindb/pull/3850/changes#r3782589607 easy would make this easy: replacing kwargs.get() with kwargs["type"] is UNSET.

I agree, this is logically imprecise. I am correcting it in upcoming changes

falexwolf · 2026-08-14T09:32:19Z

 def test_name_lookup():
    my_type = ln.Record(name="MyType", is_type=True).save()
    label1 = ln.Record(name="label 1", type=my_type).save()
+    # same type → returns existing typed record


Great idea to add the comments here! 🙏

falexwolf · 2026-08-14T09:32:48Z

+    # same type → returns existing typed record
    label2 = ln.Record(name="label 1", type=my_type)
    assert label2 == label1
+    # no type passed, only typed record exists → fallback returns the typed one


Great! This is the behavior change now.

falexwolf · 2026-08-14T09:33:44Z

@@ -690,14 +690,17 @@ def test_feature_manager_raise_not_validated_values():
 def test_name_lookup():


Aren't we now missing a test here where the user explicitly passes type=None?

falexwolf · 2026-08-14T09:35:25Z

Thanks, @ishitajain9717! This is now much better designed than yesterday! 🙏

Can you mark the comments that you resolved as resolved?

Now only very small issues are left, for which I made new comments. ☺️

falexwolf · 2026-08-14T10:31:23Z

        type = kwargs["type"]
-        if type is UNSET:
+        if type is None:
+            # explicit type=None → user wants a new root-level record, skip dedup


This is wrong, isn't it? If we pass type=None we need to deduplicate on the root level.

# explicit type=None → search root-level (type IS NULL) subset = record.__class__.filter(type__isnull=True)

I realised it immediately, and I was just correcting it 😅. I was actually reviewing changes

falexwolf · 2026-08-14T10:32:24Z

+    label_new = ln.Record(name="label 1", type=None)
+    assert label_new != label1
+    assert label_new._state.adding  # not yet saved, truly a new record
+    # explicit type=None, even if a root-level record exists → still skips dedup


ishitajain9717 added 4 commits August 12, 2026 15:45

broaden type-scoped dedup to search all contexts when type= is not pa…

2c0725c

…ssed

add fallback to search typed records when root-level dedup finds nothing

784951d

making changes so test does not break retaining originl behaviour and…

e8745e5

… fallback

solving duplication

1b60822

ishitajain9717 force-pushed the fixing_creating_projects branch from 69224bd to 1b60822 Compare August 12, 2026 13:53

ishitajain9717 added 2 commits August 12, 2026 16:09

minor changes in schema

4f7a15c

minor changes in tests

c92a784

github-actions Bot deployed to preview August 12, 2026 15:03 View deployment

falexwolf changed the title ~~Broaden type scoped deduplication to search all contexts when type= is not passed~~ 🚸 Broaden type scoped deduplication to search all contexts when type= is not passed Aug 13, 2026

falexwolf changed the title ~~🚸 Broaden type scoped deduplication to search all contexts when type= is not passed~~ 🚸 Broaden type-scoped name deduplication to search all contexts if no type is passed Aug 13, 2026

falexwolf changed the title ~~🚸 Broaden type-scoped name deduplication to search all contexts if no type is passed~~ 🚸 Broaden name deduplication in HasType to search all contexts if no type is passed Aug 13, 2026

falexwolf changed the title ~~🚸 Broaden name deduplication in HasType to search all contexts if no type is passed~~ 🚸 Broaden name deduplication in HasType to search all contexts if no type is passed Aug 13, 2026

using unset here

bc43b13

github-actions Bot deployed to preview August 13, 2026 13:41 View deployment

falexwolf reviewed Aug 13, 2026

View reviewed changes

ishitajain9717 added 2 commits August 14, 2026 08:28

removing type_val

b917409

minor change

03d36ca

github-actions Bot deployed to preview August 14, 2026 06:51 View deployment

falexwolf reviewed Aug 14, 2026

View reviewed changes

minor changed in tests and the syntax

6a01541

github-actions Bot deployed to preview August 14, 2026 10:28 View deployment

falexwolf reviewed Aug 14, 2026

View reviewed changes

minor changes in conditioning

9033d3a

github-actions Bot deployed to preview August 14, 2026 13:01 View deployment

		@@ -690,14 +690,17 @@ def test_feature_manager_raise_not_validated_values():
		def test_name_lookup():

Uh oh!

Conversation

ishitajain9717 commented Aug 10, 2026 • edited by falexwolf Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

falexwolf commented Aug 11, 2026

Uh oh!

ishitajain9717 commented Aug 11, 2026

Uh oh!

codecov Bot commented Aug 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

github-actions Bot commented Aug 12, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

falexwolf Aug 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

falexwolf commented Aug 14, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ishitajain9717 Aug 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

ishitajain9717 commented Aug 10, 2026 •

edited by falexwolf

Loading

codecov Bot commented Aug 12, 2026 •

edited

Loading

github-actions Bot commented Aug 12, 2026 •

edited

Loading

falexwolf Aug 13, 2026 •

edited

Loading

ishitajain9717 Aug 14, 2026 •

edited

Loading