Jax Liu

See the discussion https://github.com/Canner/wren-engine/pull/519#issuecomment-2041939997

# Description The Wren engine provides the deploy API for MDL deployment after the server has started. The MDL file should not be required for the server to start. Currently,...

Metabase will use `SUBSTRING` function in PostgreSQL. ``` SUBSTRING(`Model_Deal`.`company_id`, 1, 1234) `substring1189` ``` We don't transfer this to the BigQuery corresponding function correctly.

In the trino function list, `date_trunc` return `TIMESTAMP`. However, trino also supports the implicit casting for `TIMESTAMP` and `DATE` type. It can work fine to compare a `date` and `date_trunc()`...

We will create some required PG functions in data source side. Because most of them doesn't support overloading UDF or macro name, we should create different name for different function...

Because a timestamp literal or a generic literal only mention the type is `TIMESTAMP`, `ExpressionTypeAnalyzer` can't recognize if it is a `TIMESTAMP` or `TIMESTAMP_TZ`. We should dig into the value...

For now, we doesn't support to analyze the type of a [Quantified Comparison Expression](https://trino.io/docs/current/functions/comparison.html#quantified-comparison-predicates-all-any-and-some).

Some types are only defined in data warehouse. Like `datetime` is used by BigQuery but doesn't defined in the PGTypes. It's hard to analyze a field if the real type...

In `PGTypes`, we only handle the real name of a PG type. Like `INT8` is the real name of `BigIntegerType` but the most used is `bigint`. We should handle those...

`node-postgres` will pass all parameter as a VARCHAR data, no matter what type you input. ```javascript client.connect() .then(() => { const query = { text: 'SELECT * FROM Model_Ticket LIMIT...