Helper: create a database.

Helper: create a database + empty table.

Setup the `system` stats file along with its path

Consistently hash a table and namespace to a `T`. For use in a situation where you don't have a payload.

Calculate the time range that would have been stored in the shard group created when the query was prepared in InfluxQL OG ([source][1]). ```go // Mo

Check if a checkpoint should be persisted based on elapsed time.

Produce a record batch stream containing the results for the `SHOW DATABASES` query

Show retention policies for databases. This queries system.databases to show database-level retention periods.

A limited implementation of SHOW RETENTION POLICIES that assumes any database has a single, default, retention policy.

Return the approximate memory size of the chunk, in bytes including the dictionary, tables, and their rows.

Memory size in bytes including `self`

Get size of `self.columns` EXCLUDING the type itself (i.e. `size_of_val(&self.columns)`).

Slice the input batch so that it has one context row before the next input offset.

The snapshot's catalog sequence number (read from its 64-byte header via a ranged GET), or `None` if no snapshot exists.

Iterator of all of the columns in the sort key and potentially the chunk order column.

Combination of all columns within the sort key and potentially the chunk order column.

Sort groups in some order that we think is nice.

Given columns and their cardinalities (the number of distinct values in the data), sort the columns by cardinality and turn that ordering into a [`Sor

Sorts the lines (assumed to be the output of `batches_to_lines` for stable comparison)

Sort the output of this plan with the specified expressions &

Return a reference to the span contained in this SpanRecorder, or None if there is no active span

This function is to split all files in the same ParquetSource into different groups/DF partitions and set the `preserve_partitioning` so they will be

Split the RecordBatch into RecordBatches of sizes start_size, start_size + 1, ...,

Split group into sub-groups based on partition. That works because we know that -- by design -- partitions have disjunct primary key spaces.

splits input line protocol into one or more sizes of at most `max_chunk` on line breaks in a separte tokio task

This function does the actual splitting: evaluates `split_exprs` on each input [`RecordBatch`], and then sends the rows to the correct output `tx[i]`

Logically plan a SQL query against the data in a namespace, and return a DataFusion logical plan.

Physically plan a SQL query against the data in a namespace, and return a DataFusion physical execution plan.

Returns the length of time (ms) that this entry has been in the cache without being accessed. We return this value to make code using this method les

if in State::New, sets up the output running and sets self.state --> `Running`

Enables debug logging regardless of the value of RUST_LOG environment variable. If RUST_LOG isn't specifies, defaults to "debug" Hint: Try running yo

Return the stashed previous value used to fill gaps. # Panics This method will panic if `self` is not [AggrColState::PrevNullAsMissingStashed].

Return the state in which the arguments are stored

Store with a snapshot at sequence 5, a pre-snapshot log (seq 3), and a post-snapshot log (seq 7).

Returns the bytes associated by storing min/max string values

Strip `influxql::filled` metadata from all schemas in the logical plan tree. This is used before physical planning to remove InfluxQL-specific metada

Strip `influxql::filled` metadata from all fields in a schema. This is used before physical planning to remove InfluxQL-specific metadata that was us

Name of the output struct field that holds other values that just point to the selected row but for which we do NOT evaluate first/last/min/max.

Create a new context for the select statement that is a subquery of the current context.

Subscribe to the catalog for updates. This allows components in the system to listen for updates made to the catalog and handle/apply them as needed.

Branch to summary subcommand

Handles 2 scenarios: (1) when the regrouping & reordering of the `DataSourceExec` is sufficient to use ProgressiveEval (2) when the reordering of the

NOTE(tjh): A table references its parent database by id, but `TableSnapshot` doesn't store one — the id comes from the caller. `database_name` is also

NOTE(tjh): A processing-engine trigger references its parent database by id, but `ProcessingEngineTriggerSnapshot` only stores the database's name — `