Title: feat: Lance as an engine-agnostic, catalog-backed, version-pinned data source · Issue #6899 · feast-dev/feast · GitHub
Open Graph Title: feat: Lance as an engine-agnostic, catalog-backed, version-pinned data source · Issue #6899 · feast-dev/feast
X Title: feat: Lance as an engine-agnostic, catalog-backed, version-pinned data source · Issue #6899 · feast-dev/feast
Description: Summary Feast has no Lance support today — no issue or PR references it. This proposes LanceSource, but the more useful framing is that Lance is a second implementation of the abstraction already agreed in #6499, and it is the motivating...
Open Graph Description: Summary Feast has no Lance support today — no issue or PR references it. This proposes LanceSource, but the more useful framing is that Lance is a second implementation of the abstraction already a...
X Description: Summary Feast has no Lance support today — no issue or PR references it. This proposes LanceSource, but the more useful framing is that Lance is a second implementation of the abstraction already a...
Opengraph URL: https://github.com/feast-dev/feast/issues/6899
X: @github
Domain: github.com
{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"feat: Lance as an engine-agnostic, catalog-backed, version-pinned data source","articleBody":"## Summary\n\nFeast has no Lance support today — no issue or PR references it. This proposes `LanceSource`, but the more useful framing is that Lance is a **second implementation of the abstraction already agreed in #6499**, and it is the motivating use case that makes #5782, #5652 and #5330 concrete rather than theoretical.\n\nFiling it as one issue because those three are only separable on paper: an embedding source is useless without a version pin, a declared vector schema, and a choice of execution engine.\n\n## Why this is not #1494 / #1533\n\nFormat-specific sources have a poor record here — #1494 (Hudi) was closed wontfix, #1533 (Delta) has been open since 2021. So this is deliberately *not* \"add another format\".\n\nIn #6499, @ntkathole proposed moving catalog connection config onto the `DataSource` so the offline store stays generic and other engines (Trino, DuckDB) can serve the same tables; @falloficaruss agreed. That abstraction is currently specified against exactly one format and one catalog API. Lance stresses it in two ways Iceberg does not:\n\n1. **A non-Iceberg format inside a REST catalog.** Apache Polaris serves Lance through its generic-tables API, on a different path from its Iceberg API. If the source only models Iceberg tables, catalog-backed formats need a parallel code path — cheaper to settle before `IcebergRestCatalogSource` lands than after.\n2. **A non-JVM read path.** Lance has a native Python/Rust reader, so this source can be served with no Spark at all. That is a real test of whether the `DataSource` is engine-agnostic, rather than Spark-agnostic in name only.\n\n## Proposal\n\n### `LanceSource`\n\n```python\nLanceSource(\n catalog=\"my_catalog\", # catalog-backed\n namespace=\"my_namespace\",\n table=\"item_embeddings\",\n # or path-based:\n # uri=\"s3://bucket/item_embeddings.lance\",\n\n version=3, # or tag=\"candidate\" — #5782\n timestamp_field=\"generated_at\",\n)\n```\n\nA naming question worth settling early: if #6499 lands as `IcebergRestCatalogSource`, then a sibling called `LanceSource` mixes two axes — one named for a catalog, one for a format. Either the catalog connection is a shared, reusable piece that both formats reference, or the names should agree on an axis. I don't have a strong preference, but it is easier to decide now than to rename later.\n\nVector schema is declared on the FeatureView, using fields `Field` already has:\n\n```python\nField(name=\"embedding\", dtype=Array(Float32),\n vector_index=True, vector_length=160, vector_search_metric=\"COSINE\")\n```\n\n### Two engines, one source — #5330\n\n| Path | Engine | Use |\n|---|---|---|\n| Native | `pylance` + DataFusion/DuckDB | local dev, CI, notebooks, eval-set construction |\n| Spark | `lance-spark` | existing batch/GPU clusters |\n\nThe read-path half is already solved: `HybridOfflineStore` (#5541, shipped in 0.51.0) routes by data source type, so both stores can coexist in one project.\n\nThe remaining gap is #5330 — `batch_engine` / `batch_configs` on `FeatureView` are still placeholders, so a project cannot choose the engine *per FeatureView*. An embedding pipeline needs exactly that: Spark for generation, native for interactive retrieval, one project. #6359 looks like the structural prerequisite, since the offline-store retrieval path currently bypasses the compute engine.\n\nPrecedent for one source across engines: `FileSource` is served by both the Dask and DuckDB offline stores.\n\n### Pin semantics — answering @jfw-ppi on #5782\n\n@jfw-ppi asked whether the response schema would stay FeatureView-defined rather than varying by store or snapshot. Proposed:\n\n- The `FeatureView` schema is the contract. A pinned version selects **data**, never shape.\n- At resolve time, validate the pinned version's actual schema against the declared schema.\n- Incompatibility **fails explicitly**. A vector-length change is a hard error, not a silently different response.\n\nRelated, and an argument for validating rather than trusting declarations: in `Field.__eq__` the `vector_index` and `vector_search_metric` comparisons are currently commented out, so only `vector_length` participates in equality:\n\n```python\nor self.vector_length != other.vector_length\n# or self.vector_index != other.vector_index\n# or self.vector_search_metric != other.vector_search_metric\n```\n\nA metric change from `COSINE` to `L2` is therefore invisible to `feast apply` / plan. Happy to split that out into its own bug if preferred.\n\n### Vector config placement — #5652\n\nThis assumes #5652's direction: vector config belongs to the feature definition, not global `online_store` config. The `Field` attributes already exist, but online stores still read dimension and metric from global config, so two feature views with different dimensions cannot coexist in one project.\n\n## Scope\n\n**Not** proposing Lance as an online store. Lance is a format plus indexes, not a low-latency KV service, and its write path is columnar while `online_write_batch` is row-oriented proto. Offline / retrieval axis only.\n\n## Suggested sequencing\n\n1. `LanceSource` + native (pylance) offline store, path-based only — landable without waiting on #6499\n2. Catalog-backed addressing, coordinated with #6499\n3. `version` / `tag` pinning + schema validation (#5782)\n4. Spark offline store support for `LanceSource`\n5. Per-FeatureView engine selection (#5330, after #6359)\n\nRelated: #6499, #5782, #5652, #5330, #6359, #5541, #2406\n","author":{"url":"https://github.com/haoxu0","@type":"Person","name":"haoxu0"},"datePublished":"2026-09-29T23:28:09Z","url":"https://github.com/feast-dev/feast/issues/6899"}
| react-profiling | 0 |
| react-import-map | react,react-dom,react-dom/client,react-dom/profiling,react-is,react/compiler-runtime,react/jsx-dev-runtime,react/jsx-runtime,scheduler@b8488ee2c322 |
| route-pattern | /_view_fragments/issues/show/:user_id/:repository/:id/issue_layout(.:format) |
| route-controller | voltron_issues_fragments |
| route-action | issue_layout |
| fetch-nonce | v2:487e519b-db82-2b40-56d7-244edc13a5ea |
| current-catalog-service-hash | 81bb79d38c15960b92d99bca9288a9108c7a47b18f2423d0f6438c5b7bcd2114 |
| request-id | 9D7C:37D9F:F2D953:14C4F7A:6AFE9810 |
| html-safe-nonce | 830ac087ba1845dc526f3f0b549544343ead842375c98f0f5eb4aa5542d0c21b |
| visitor-payload | eyJyZWZlcnJlciI6IiIsInJlcXVlc3RfaWQiOiI5RDdDOjM3RDlGOkYyRDk1MzoxNEM0RjdBOjZBRkU5ODEwIiwidmlzaXRvcl9pZCI6Ijg0OTU4MjEzOTM1MDQ0MDk2MTYiLCJyZWdpb25fZWRnZSI6ImlhZCIsInJlZ2lvbl9yZW5kZXIiOiJpYWQifQ== |
| visitor-hmac | 84bc6cde570a7341e7199e0a6f6cfb230a218f698bd326b9010cb5371b674556 |
| hovercard-subject-tag | issue:5638466360 |
| github-keyboard-shortcuts | repository,issues,copilot |
| google-site-verification | Apib7-x98H0j5cPqHWwSMm6dNU4GmODRoqxLiDzdx9I |
| octolytics-url | https://collector.github.com/github/collect |
| analytics-location | / |
| fb:app_id | 1401488693436528 |
| apple-itunes-app | app-id=1477376905, app-argument=https://github.com/_view_fragments/issues/show/feast-dev/feast/6899/issue_layout |
| twitter:image | https://opengraph.githubassets.com/1474230da1612f7a337d126f5d42ee74c4b8ab055fac7e5501cd0f9a5b903074/feast-dev/feast/issues/6899 |
| twitter:card | summary_large_image |
| og:image | https://opengraph.githubassets.com/1474230da1612f7a337d126f5d42ee74c4b8ab055fac7e5501cd0f9a5b903074/feast-dev/feast/issues/6899 |
| og:image:alt | Summary Feast has no Lance support today — no issue or PR references it. This proposes LanceSource, but the more useful framing is that Lance is a second implementation of the abstraction already a... |
| og:image:width | 1200 |
| og:image:height | 600 |
| og:site_name | GitHub |
| og:type | object |
| og:author:username | haoxu0 |
| hostname | github.com |
| expected-hostname | github.com |
| None | b3c0f155eafc27e60db6e254d1ffff0e072e9ab3b4018a5ee518797c19d89f86 |
| turbo-cache-control | no-preview |
| go-import | github.com/feast-dev/feast git https://github.com/feast-dev/feast.git |
| octolytics-dimension-user_id | 57027613 |
| octolytics-dimension-user_login | feast-dev |
| octolytics-dimension-repository_id | 161133770 |
| octolytics-dimension-repository_nwo | feast-dev/feast |
| octolytics-dimension-repository_public | true |
| octolytics-dimension-repository_is_fork | false |
| octolytics-dimension-repository_network_root_id | 161133770 |
| octolytics-dimension-repository_network_root_nwo | feast-dev/feast |
| turbo-body-classes | logged-out env-production page-responsive |
| disable-turbo | false |
| browser-stats-url | https://api.github.com/_private/browser/stats |
| browser-errors-url | https://api.github.com/_private/browser/errors |
| release | 4a10310af51a1bfffbedc4fd81cb582acf0cb0ff |
| ui-target | full |
| theme-color | #1e2327 |
| color-scheme | light dark |
| react-app-name | issues |
Links:
Viewport: width=device-width