Make last updated timestamp robust

[abey79]

Related

• part of https://linear.app/rerun/issue/RR-2640/repeated-dataframe-query-view-calls-slow-on-specific-rrd-for-oss
• part of https://linear.app/rerun/issue/RR-2715/registering-with-layers-does-not-do-an-update-to-the-dataset-updated
• sibling https://github.com/rerun-io/dataplatform/pull/1985

What

I want to introduce a caching mechanism for dataset schema, since this operation is currently very expansive and often used in dataframe queries. A simple way to invalidate that cache would be to use the last_updated_at field that we maintain anyways. As a first step, this PR makes it more robust to maintain that last_update_at flag by introducing a Tracked helper which does that automagically.

Also introduces a couple of tests.

[github-actions]

Web viewer built successfully.

Result	Commit	Link	Manifest
✅	545bbe9	https://rerun.io/viewer/pr/11807	+nightly +main

View image diff on kitdiff.

^{Note: This comment is updated whenever you push a commit.}

[Copilot]

Pull Request Overview

This PR introduces a Tracked<T> wrapper type to automatically maintain last_updated_at timestamps for dataset state mutations. The wrapper provides immutable access via Deref and mutable access via a guard pattern that updates timestamps on drop. The implementation refactors Dataset and Partition to use Tracked for their mutable state, ensuring timestamp updates occur automatically whenever state is modified.

Key changes:

• New Tracked<T> type with guard-based mutation tracking
• Refactored Dataset and Partition to wrap mutable state in Tracked
• Added tests verifying timestamp updates on rename and partition registration operations

Reviewed Changes

Copilot reviewed 8 out of 9 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
crates/store/re_server/src/store/tracked.rs	Introduces the new Tracked<T> wrapper and TrackedGuard types for automatic timestamp tracking
crates/store/re_server/src/store/partition.rs	Refactors Partition to use Tracked<PartitionInner> for automatic timestamp updates
crates/store/re_server/src/store/mod.rs	Exports the new Tracked type from the store module
crates/store/re_server/src/store/dataset.rs	Refactors Dataset to use Tracked<DatasetInner> and adds updated_at() accessor
crates/store/re_redap_tests/src/tests/update_entry.rs	Adds test verifying timestamp updates when renaming datasets
crates/store/re_redap_tests/src/tests/register_partition.rs	Adds test verifying timestamp updates when registering partitions and layers
crates/store/re_redap_tests/src/tests/mod.rs	Registers the new timestamp-related tests
crates/store/re_redap_tests/Cargo.toml	Adds tokio dependency for async test delays

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[oxkitsune]

very elegant and satisfying solution!

[oxkitsune]

Similarly to the set_name shouldn't it be trivial to check whether the new details are actually new? In the case where they are, this will update the last_updated_timestamp even though nothing changed.

DatasetDetails can implement PartialEq and Eq, all its components already do.