Repository navigation
Real-server tests, streaming and server-side Stop for more engines, plans for document engines - #312
Merged
Merged
Conversation
…erver sessions Transactions round-trip on real SQL Server and Oracle servers (rollback, commit, abandoned and released). ClickHouse streams 2M rows and Stop leaves nothing in system.processes. A transaction's SQL Server client now leaves the cancel registry when it closes instead of keeping its login there.
Oracle queries feed the grid or result store as the session fetches them instead of collecting up to the budget; Stop breaks the call. MongoDB find and aggregate stream from the cursor past the console's 5,000-document cap, with fields the first 1,000 documents lacked kept in (other fields). Both verified on real servers.
Elasticsearch and OpenSearch requests carry an X-Opaque-Id and Stop cancels the tasks holding it. PostHog queries carry a client_query_id and Stop sends DELETE for it. A streamed MongoDB find or aggregate is tagged with a comment and Stop runs killOp on it. Elasticsearch and MongoDB verified on real servers; PostHog against a fake that checks the cancel call.
Explain draws MongoDB's executed stages (aggregate stages above the cursor), a search's profile per shard and Lucene query, CouchDB's _explain index choice, and libSQL's SQLite plan in the plan view. The full-scan warning knows COLLSCAN and CouchDB's all-documents scan. Verified on real MongoDB 7, Elasticsearch 8 and CouchDB 3.
Menu and popover filter rows (chart picker, database switcher, data diff pickers, column finder, model picker) are now an inset search-shell, the same outline as every page search. The terminal's find box drops its drop shadow for the outline and lights its border on focus.
Console reads execute in DuckDB's streaming mode and are read through the row API (raw_query), so DuckDB no longer materializes the whole result first: 2M rows now grow memory under 120 MB where DuckDB alone held 214 MB. The driver's Arrow iterator panics on a mid-stream error; the row API returns it, and a test covers that.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The second round on the five hard cases. This time I tested against real servers (SQL Server, Oracle 23, ClickHouse, MongoDB 7, Elasticsearch 8, CouchDB 3), not only fakes.
Big results
findandaggregatestream from the cursor past the old 5,000-document cap. Fields that first appear after the first 1,000 documents are kept in(other fields).Stop
X-Opaque-Id, and Stop cancels the tasks holding it.client_query_id, and Stop sends DELETE for it.killOpon them.Plans
_explain) and libSQL.Transactions
UI
Startup
Checks: