Allow tables to be created without providing a URL

crates/store/re_protos/proto/rerun/v1alpha1/cloud.proto

@@ -862,9 +862,9 @@ message CreateTableEntryRequest {
 
   // Information about the table to register.
   //
-  // This must be encoded message of one one of the following supported types:
+  // This must be encoded message of one of the following supported types:
   // - LanceTable
-  google.protobuf.Any provider_details = 2;
+  optional google.protobuf.Any provider_details = 2;
 
   // Schema of the table to create
   rerun.common.v1alpha1.Schema schema = 3;

crates/store/re_protos/src/v1alpha1/rerun.cloud.v1alpha1.ext.rs

@@ -775,7 +775,7 @@ impl TryFrom<crate::cloud::v1alpha1::CreateDatasetEntryResponse> for CreateDatas
 pub struct CreateTableEntryRequest {
     pub name: String,
     pub schema: Schema,
-    pub provider_details: ProviderDetails,
+    pub provider_details: Option<ProviderDetails>,
 }
 
 impl TryFrom<CreateTableEntryRequest> for crate::cloud::v1alpha1::CreateTableEntryRequest {
@@ -784,7 +784,10 @@ impl TryFrom<CreateTableEntryRequest> for crate::cloud::v1alpha1::CreateTableEnt
         Ok(Self {
             name: value.name,
             schema: Some((&value.schema).try_into()?),
-            provider_details: Some((&value.provider_details).try_into()?),
+            provider_details: value
+                .provider_details
+                .map(|d| (&d).try_into())
+                .transpose()?,
         })
     }
 }
@@ -803,12 +806,10 @@ impl TryFrom<crate::cloud::v1alpha1::CreateTableEntryRequest> for CreateTableEnt
                     "schema"
                 ))?
                 .try_into()?,
-            provider_details: ProviderDetails::try_from(&value.provider_details.ok_or(
-                missing_field!(
-                    crate::cloud::v1alpha1::CreateTableEntryRequest,
-                    "provider_details"
-                ),
-            )?)?,
+            provider_details: value
+                .provider_details
+                .map(|v| ProviderDetails::try_from(&v))
+                .transpose()?,
         })
     }
 }

crates/store/re_redap_client/src/connection_client.rs

@@ -837,12 +837,11 @@ where
     pub async fn create_table_entry(
         &mut self,
         name: &str,
-        url: &Url,
+        url: Option<Url>,
         schema: SchemaRef,
     ) -> ApiResult<TableEntry> {
-        let provider_details = ProviderDetails::LanceTable(LanceTable {
-            table_url: url.clone(),
-        });
+        let provider_details =
+            url.map(|url| ProviderDetails::LanceTable(LanceTable { table_url: url }));
         let request = CreateTableEntryRequest {
             name: name.to_owned(),
             schema: schema.as_ref().clone(),

crates/store/re_redap_tests/src/tests/create_dataset.rs

@@ -144,7 +144,7 @@ pub async fn create_dataset_tests(service: impl RerunCloudService) {
     let create_table_request = CreateTableEntryRequest {
         name: table_name.to_owned(),
         schema: schema.clone(),
-        provider_details,
+        provider_details: Some(provider_details),
     }
     .try_into()
     .expect("Unable to create table request");

crates/store/re_redap_tests/src/tests/create_table.rs

@@ -28,7 +28,7 @@ pub async fn create_table_entry(service: impl RerunCloudService) {
     let create_table_request = CreateTableEntryRequest {
         name: table_name.to_owned(),
         schema: schema.clone(),
-        provider_details,
+        provider_details: Some(provider_details),
     }
     .try_into()
     .expect("Unable to create table request");

crates/store/re_redap_tests/src/tests/update_entry.rs

@@ -275,7 +275,7 @@ async fn create_table_entry(
             CreateTableEntryRequest {
                 name: table_name.to_owned(),
                 schema: schema.clone(),
-                provider_details,
+                provider_details: Some(provider_details),
             }
             .try_into()
             .unwrap(),

crates/store/re_server/src/rerun_cloud.rs

@@ -15,6 +15,7 @@ use re_arrow_util::RecordBatchExt as _;
 use re_chunk_store::{Chunk, ChunkStore, ChunkStoreHandle};
 use re_log_encoding::ToTransport as _;
 use re_log_types::{EntityPath, EntryId, StoreId, StoreKind};
+use re_protos::cloud::v1alpha1::ext::LanceTable;
 use re_protos::cloud::v1alpha1::ext::{
     self, CreateDatasetEntryRequest, CreateDatasetEntryResponse, CreateTableEntryRequest,
     CreateTableEntryResponse, DataSource, EntryDetailsUpdate, ProviderDetails, QueryDatasetRequest,
@@ -39,14 +40,28 @@ use tokio_stream::StreamExt as _;
 use tonic::{Code, Request, Response, Status};
 
 use crate::OnError;
+use re_tuid::Tuid;
 
-#[derive(Debug, Default)]
-pub struct RerunCloudHandlerSettings {}
+#[derive(Debug)]
+pub struct RerunCloudHandlerSettings {
+    storage_dir: tempfile::TempDir,
+}
+
+impl Default for RerunCloudHandlerSettings {
+    fn default() -> Self {
+        Self {
+            storage_dir: create_data_dir().expect("Failed to create data directory"),
+        }
+    }
+}
+
+fn create_data_dir() -> Result<tempfile::TempDir, crate::store::Error> {
+    Ok(tempfile::Builder::new().prefix("rerun-data-").tempdir()?)
+}
 
 #[derive(Default)]
 pub struct RerunCloudHandlerBuilder {
     settings: RerunCloudHandlerSettings,
-
     store: InMemoryStore,
 }
 
@@ -119,7 +134,6 @@ impl RerunCloudHandlerBuilder {
 const DUMMY_TASK_ID: &str = "task_00000000DEADBEEF";
 
 pub struct RerunCloudHandler {
-    #[expect(dead_code)]
     settings: RerunCloudHandlerSettings,
 
     store: tokio::sync::RwLock<InMemoryStore>,
@@ -1509,7 +1523,27 @@ impl RerunCloudService for RerunCloudHandler {
 
         let schema = Arc::new(request.schema);
 
-        let table = match &request.provider_details {
+        let details = if let Some(details) = request.provider_details {
+            details
+        } else {
+            // Create a directory in the storage directory. We use a tuid to avoid collisions
+            // and avoid any sanitization issue with the provided table name.
+            let table_path = self
+                .settings
+                .storage_dir
+                .path()
+                .join(format!("lance-{}", Tuid::new()));
+            ProviderDetails::LanceTable(LanceTable {
+                table_url: url::Url::from_directory_path(table_path).map_err(|_err| {
+                    Status::internal(format!(
+                        "Failed to create table directory in {:?}",
+                        self.settings.storage_dir.path()
+                    ))
+                })?,
+            })
+        };
+
+        let table = match details {
             ProviderDetails::LanceTable(table) => {
                 store
                     .create_table_entry(table_name, &table.table_url, schema)

rerun_py/rerun_bindings/rerun_bindings.pyi

@@ -1513,7 +1513,7 @@ class CatalogClientInternal:
 
     def create_dataset(self, name: str) -> DatasetEntryInternal: ...
     def register_table(self, name: str, url: str) -> TableEntryInternal: ...
-    def create_table(self, name: str, schema: pa.Schema, url: str) -> TableEntryInternal: ...
+    def create_table(self, name: str, schema: pa.Schema, url: str | None) -> TableEntryInternal: ...
     def ctx(self) -> dfn.SessionContext: ...
 
     # ---

rerun_py/rerun_sdk/rerun/catalog/_catalog_client.py

@@ -290,7 +290,7 @@ def register_table(self, name: str, url: str) -> TableEntry:
 
         return TableEntry(self._internal.register_table(name, url))
 
-    def create_table(self, name: str, schema: pa.Schema, url: str) -> TableEntry:
+    def create_table(self, name: str, schema: pa.Schema, url: str | None = None) -> TableEntry:
         """
         Create and register a new table.
 

rerun_py/src/catalog/catalog_client.rs

@@ -238,7 +238,7 @@ impl PyCatalogClientInternal {
         py: Python<'_>,
         name: String,
         schema: PyArrowType<Schema>,
-        url: String,
+        url: Option<String>,
     ) -> PyResult<Py<PyTableEntryInternal>> {
         let connection = self_.borrow_mut(py).connection.clone();
 
@@ -250,11 +250,14 @@ impl PyCatalogClientInternal {
             .map_err(|err| PyValueError::new_err(format!("Invalid table name. {err}")))?;
 
         let url = url
-            .parse::<url::Url>()
-            .map_err(|err| PyValueError::new_err(format!("Invalid URL: {err}")))?;
+            .map(|url| {
+                url.parse::<url::Url>()
+                    .map_err(|err| PyValueError::new_err(format!("Invalid URL: {err}")))
+            })
+            .transpose()?;
 
         let schema = Arc::new(schema.0);
-        let table_entry = connection.create_table_entry(py, name, schema, &url)?;
+        let table_entry = connection.create_table_entry(py, name, schema, url)?;
 
         self_.borrow(py).update_catalog_providers(py, false)?;
 

rerun_py/src/catalog/connection_handle.rs

@@ -209,7 +209,7 @@ impl ConnectionHandle {
         py: Python<'_>,
         name: String,
         schema: SchemaRef,
-        url: &url::Url,
+        url: Option<url::Url>,
     ) -> PyResult<TableEntry> {
         let entry_id = wait_for_future(
             py,

rerun_py/tests/e2e_redap_tests/conftest.py

@@ -206,7 +206,7 @@ def create_dataset(self, name: str) -> DatasetEntry:
         self._created_entries.append(entry)
         return entry
 
-    def create_table(self, name: str, schema: pa.Schema, url: str) -> TableEntry:
+    def create_table(self, name: str, schema: pa.Schema, url: str | None = None) -> TableEntry:
         """Create a table with automatic cleanup. Mirrors CatalogClient.create_table()."""
         prefixed_name = self.apply_prefix(name)
         entry = self._client.create_table(prefixed_name, schema, url)

rerun_py/tests/e2e_redap_tests/test_entries.py

@@ -6,15 +6,13 @@
 import pytest
 
 if TYPE_CHECKING:
-    from pathlib import Path
-
     from syrupy import SnapshotAssertion
 
     from e2e_redap_tests.conftest import EntryFactory
 
 
 @pytest.mark.creates_table
-def test_entries_without_hidden(entry_factory: EntryFactory, tmp_path: Path, snapshot: SnapshotAssertion) -> None:
+def test_entries_without_hidden(entry_factory: EntryFactory, snapshot: SnapshotAssertion) -> None:
     """Test that entries(), datasets(), and tables() exclude hidden entries by default."""
     client = entry_factory.client
 
@@ -25,7 +23,7 @@ def test_entries_without_hidden(entry_factory: EntryFactory, tmp_path: Path, sna
 
     # Create test entries
     entry_factory.create_dataset("test_dataset")
-    entry_factory.create_table("test_table", pa.schema([pa.field("col", pa.int32())]), tmp_path.as_uri())
+    entry_factory.create_table("test_table", pa.schema([pa.field("col", pa.int32())]))
 
     # Get entries after - should only show user-created entries (no hidden)
     datasets_after = {d.name for d in client.datasets()}
@@ -44,9 +42,7 @@ def test_entries_without_hidden(entry_factory: EntryFactory, tmp_path: Path, sna
 
 
 @pytest.mark.creates_table
-def test_entries_with_hidden(
-    entry_factory: EntryFactory, tmp_path: Path, snapshot_redact_id: SnapshotAssertion
-) -> None:
+def test_entries_with_hidden(entry_factory: EntryFactory, snapshot_redact_id: SnapshotAssertion) -> None:
     """Test that entries(), datasets(), and tables() include hidden entries when include_hidden=True."""
     client = entry_factory.client
 
@@ -57,7 +53,7 @@ def test_entries_with_hidden(
 
     # Create test entries
     entry_factory.create_dataset("test_dataset")
-    entry_factory.create_table("test_table", pa.schema([pa.field("col", pa.int32())]), tmp_path.as_uri())
+    entry_factory.create_table("test_table", pa.schema([pa.field("col", pa.int32())]))
 
     # Get entries after with hidden - should include blueprint datasets and system tables
     datasets_after = {d.name for d in client.datasets(include_hidden=True)}
@@ -76,7 +72,7 @@ def test_entries_with_hidden(
 
 
 @pytest.mark.creates_table
-def test_entry_names_without_hidden(entry_factory: EntryFactory, tmp_path: Path, snapshot: SnapshotAssertion) -> None:
+def test_entry_names_without_hidden(entry_factory: EntryFactory, snapshot: SnapshotAssertion) -> None:
     """Test that entry_names(), dataset_names(), and table_names() exclude hidden entries by default."""
     client = entry_factory.client
 
@@ -87,7 +83,7 @@ def test_entry_names_without_hidden(entry_factory: EntryFactory, tmp_path: Path,
 
     # Create test entries
     entry_factory.create_dataset("test_dataset")
-    entry_factory.create_table("test_table", pa.schema([pa.field("col", pa.int32())]), tmp_path.as_uri())
+    entry_factory.create_table("test_table", pa.schema([pa.field("col", pa.int32())]))
 
     # Get names after - should only show user-created entries (no hidden)
     dataset_names_after = set(client.dataset_names())
@@ -106,9 +102,7 @@ def test_entry_names_without_hidden(entry_factory: EntryFactory, tmp_path: Path,
 
 
 @pytest.mark.creates_table
-def test_entry_names_with_hidden(
-    entry_factory: EntryFactory, tmp_path: Path, snapshot_redact_id: SnapshotAssertion
-) -> None:
+def test_entry_names_with_hidden(entry_factory: EntryFactory, snapshot_redact_id: SnapshotAssertion) -> None:
     """Test that entry_names(), dataset_names(), and table_names() include hidden entries when include_hidden=True."""
     client = entry_factory.client
 
@@ -119,7 +113,7 @@ def test_entry_names_with_hidden(
 
     # Create test entries
     entry_factory.create_dataset("test_dataset")
-    entry_factory.create_table("test_table", pa.schema([pa.field("col", pa.int32())]), tmp_path.as_uri())
+    entry_factory.create_table("test_table", pa.schema([pa.field("col", pa.int32())]))
 
     # Get names after with hidden - should include blueprint datasets and system tables
     dataset_names_after = set(client.dataset_names(include_hidden=True))