SnowConvert AI - Redshift Issues¶

Description¶

The SET configuration parameter clause in Redshift procedures is not supported in Snowflake. Snowflake uses ALTER SESSION SET or session-level parameters instead. For more information, refer to CREATE PROCEDURE documentation.

Code Example¶

 CREATE OR REPLACE PROCEDURE procedure2(
    IN input_param INTEGER,
    OUT output_param NUMERIC
)
AS $$
BEGIN
    output_param := input_param * 1.7;
END;
$$
LANGUAGE plpgsql
SET enable_numeric_rounding to ON;

 CREATE OR REPLACE PROCEDURE procedure2 (input_param INTEGER, output_param OUT NUMERIC)
RETURNS VARCHAR
LANGUAGE SQL
!!!RESOLVE EWI!!! /*** SSC-EWI-RS0002 - SET CONFIGURATION PARAMETER 'enable_numeric_rounding' IS NOT SUPPORTED IN SNOWFLAKE. ***/!!!
SET enable_numeric_rounding to ON
COMMENT = '{ "origin": "sf_sc", "name": "snowconvert", "version": {  "major": 0,  "minor": 0,  "patch": "0" }, "attributes": {  "component": "redshift",  "convertedOn": "07/16/2025",  "domain": "no-domain-provided" }}'
AS $$
BEGIN
    output_param := input_param * 1.7;
END;
$$;

Best Practices¶

• Use ALTER SESSION SET: Snowflake provides ALTER SESSION SET to configure session-level parameters. Review whether the Redshift configuration parameter has an equivalent Snowflake session parameter.

• Remove if unnecessary: Some Redshift configuration parameters (e.g., enable_numeric_rounding) have no Snowflake equivalent and may be safely removed if Snowflake’s default behavior meets your requirements.

• If you need more support, you can email us at snowconvert-support@snowflake.com

Description¶

Redshift documentation for CREATE VIEW includes an optional clause that specifies that the particular view is not bound to the database objects such as tables or functions, nor to those objects that it is referencing. The documentation also clarifies that in such cases that this clause is used, the referenced objects must be qualified with a schema name. This clause allows to create a view and reference objects that might not exist yet. Their existence will be verified once the view is queried, but not at its definition.

However, there is no equivalent command nor obvious workaround to implement this functionality in Snowflake, furthermore, the Snowflake documentation suggests that the views are linked to a specific schema and so are the referenced objects in the view.

If the references linked to the View are present in the input code, the statement will be removed without issue. However, if the necessary references are missing, a warning message will be added to inform the user that the statement cannot be removed due to the missing references.

SnowConvert AI performs analysis solely on the input code and does not account for objects already deployed in Snowflake. Therefore the output may have some issues pointing to missing references, if the references are already present in the Snowflake database, the user can safely remove the statement without any issues.

Code Examples¶

 CREATE VIEW myView AS SELECT col1 FROM public.missingTable
WITH NO SCHEMA BINDING;

 --** SSC-FDM-0007 - MISSING DEPENDENT OBJECT "public.missingTable" **
CREATE VIEW myView
COMMENT = '{ "origin": "sf_sc", "name": "snowconvert", "version": {  "major": 0,  "minor": 0,  "patch": "0" }, "attributes": {  "component": "redshift",  "convertedOn": "09/17/2024" }}'
AS SELECT col1 FROM
public.missingTable
!!!RESOLVE EWI!!! /*** SSC-EWI-RS0003 - WITH NO SCHEMA BINDING STATEMENT CAN NOT BE REMOVED DUE TO MISSING REFERENCES. ***/!!!
WITH NO SCHEMA BINDING;

Best Practices¶

• To resolve this issue, it is suggested to add the missing references to the input code, if the object is already deployed in the Snowflake database, the statement can be remove without issue.

• If you need more support, you can email us at snowconvert-support@snowflake.com

Description¶

This conversion issue is added because the HLLSKETCH data type is not supported in Snowflake.

Code Example¶

 CREATE TABLE table1
(
    col_hllsketch HLLSKETCH
);

 CREATE TABLE table1
(
    col_hllsketch HLLSKETCH !!!RESOLVE EWI!!! /*** SSC-EWI-RS0004 - HLLSKETCH DATA TYPE NOT SUPPORTED IN SNOWFLAKE. ***/!!!
)
COMMENT = '{ "origin": "sf_sc", "name": "snowconvert", "version": {  "major": 0,  "minor": 0,  "patch": "0" }, "attributes": {  "component": "redshift",  "convertedOn": "09/17/2024" }}';

Best Practices¶

• Please verify all aggregate functions provided by Snowflake to estimate cardinality using HyperLogLog.

• If you need more support, you can email us at snowconvert-support@snowflake.com

Description¶

Pending SnowConvert AI translation for column aliases in the PIVOT/UNPIVOT IN clause. Snowflake now supports the AS clause for specifying column aliases in PIVOT and UNPIVOT operations (added October 2025). This is a SnowConvert AI limitation, not a Snowflake platform limitation.

Code Example¶

 SELECT *
FROM count_by_color UNPIVOT (
    cnt FOR color IN (red AS r, green AS g, blue AS b)
);

 SELECT *
FROM
    count_by_color UNPIVOT (
    cnt FOR color IN (red
                          !!!RESOLVE EWI!!! /*** SSC-EWI-RS0005 - PENDING SNOWCONVERT AI TRANSLATION FOR COLUMN ALIASES IN THE PIVOT/UNPIVOT IN CLAUSE. ***/!!! AS r, green
                                                                                                                                                                              !!!RESOLVE EWI!!! /*** SSC-EWI-RS0005 - PENDING SNOWCONVERT AI TRANSLATION FOR COLUMN ALIASES IN THE PIVOT/UNPIVOT IN CLAUSE. ***/!!! AS g, blue
                                                                                                                                                                                                                                                                                                                                 !!!RESOLVE EWI!!! /*** SSC-EWI-RS0005 - PENDING SNOWCONVERT AI TRANSLATION FOR COLUMN ALIASES IN THE PIVOT/UNPIVOT IN CLAUSE. ***/!!! AS b)
);

Best Practices¶

• Use native Snowflake support: Snowflake now supports the AS clause for column aliases in PIVOT/UNPIVOT IN clauses. Remove the EWI marker and use the aliases directly:

 SELECT *
FROM count_by_color UNPIVOT (
    cnt FOR color IN (red AS r, green AS g, blue AS b)
);

• Further reading: Snowflake UNPIVOT, Snowflake PIVOT

• If you need more support, you can email us at snowconvert-support@snowflake.com

Description¶

The behavior of the SUBSTRING function on binary data differs between Redshift and Snowflake. In Redshift, SUBSTRING on VARBYTE operates on raw bytes. In Snowflake, SUBSTRING on BINARY operates on hex-encoded character pairs, so the same positional arguments may return different results.

Code Example¶

 SELECT SUBSTRING('12345'::varbyte, 2, 4) AS substring_binary;
SELECT SUBSTRING('abc'::varbyte, 2, 4) AS substring_binary;

 SELECT SUBSTRING('12345':: BINARY, 2, 4) !!!RESOLVE EWI!!! /*** SSC-EWI-RS0006 - THE BEHAVIOR OF THE SUBSTRING FUNCTION ON BINARY DATA DIFFERS BETWEEN REDSHIFT AND SNOWFLAKE. ***/!!! AS substring_binary;
SELECT SUBSTRING('abc':: BINARY, 2, 4) !!!RESOLVE EWI!!! /*** SSC-EWI-RS0006 - THE BEHAVIOR OF THE SUBSTRING FUNCTION ON BINARY DATA DIFFERS BETWEEN REDSHIFT AND SNOWFLAKE. ***/!!! AS substring_binary;

Best Practices¶

• Verify binary output: Compare SUBSTRING results on binary columns between Redshift and Snowflake to confirm correctness after migration.

• Adjust offsets: Because Snowflake’s BINARY type uses hex encoding, you may need to multiply position and length arguments by 2 to achieve equivalent byte-level extraction.

• If you need more support, you can email us at snowconvert-support@snowflake.com

Description¶

Some DATE, TIME, or TIMESTAMP literal formats used in Redshift (e.g., '2000-Jan-31', 'Jan-31-2000') are not recognized by Snowflake. These literals must be rewritten to a supported Snowflake date format or converted using TO_DATE with an explicit format string.

Code Example¶

 select datediff(century, '2000-Jan-31', 'Jan-31-2000');

  select
 DATEDIFF(YEAR,
                !!!RESOLVE EWI!!! /*** SSC-EWI-RS0007 - '2000-Jan-31' DATE LITERAL IS NOT SUPPORTED IN SNOWFLAKE. ***/!!!
                '2000-Jan-31',
                               !!!RESOLVE EWI!!! /*** SSC-EWI-RS0007 - 'Jan-31-2000' DATE LITERAL IS NOT SUPPORTED IN SNOWFLAKE. ***/!!!
                               'Jan-31-2000') / 100;

Best Practices¶

• Use ISO 8601 format: Rewrite date literals to 'YYYY-MM-DD' format, which is universally supported in Snowflake.

• Use TO_DATE with format string: If the original format must be preserved, use TO_DATE('Jan-31-2000', 'MON-DD-YYYY') to explicitly parse the date.

• If you need more support, you can email us at snowconvert-support@snowflake.com

Description¶

In Redshift, you can apply the DELETE statement to materialized views used for streaming ingestion. In Snowflake, materialized views are transformed into dynamic tables, and the DELETE statement cannot be used on dynamic tables.

Code Example¶

 CREATE MATERIALIZED VIEW mv AS
SELECT id, name, department_id FROM employees WHERE department_id = 101;

DELETE FROM mv
WHERE id = 2;

 CREATE DYNAMIC TABLE mv
--** SSC-FDM-0031 - DYNAMIC TABLE REQUIRED PARAMETERS SET BY DEFAULT **
TARGET_LAG='1 day'
WAREHOUSE=UPDATE_DUMMY_WAREHOUSE
COMMENT = '{ "origin": "sf_sc", "name": "snowconvert", "version": {  "major": 0,  "minor": 0,  "patch": "0" }, "attributes": {  "component": "redshift",  "convertedOn": "03/03/2025",  "domain": "test" }}'
AS
SELECT id, name, department_id FROM
employees
WHERE department_id = 101;


!!!RESOLVE EWI!!! /*** SSC-EWI-RS0008 - MATERIALIZED VIEW IS TRANSFORMED INTO A DYNAMIC TABLE, AND THE DELETE STATEMENT CANNOT BE USED ON DYNAMIC TABLES IN SNOWFLAKE. ***/!!!
DELETE FROM
mv
WHERE id = 2;

Best Practices¶

• Replace the dynamic table definition: Because dynamic tables cannot be directly deleted from, you can achieve the same result by altering the dynamic table’s underlying query to exclude the rows you want to remove.

• Use a regular table: If row-level DML (INSERT, UPDATE, DELETE) is required, consider using a regular table with a scheduled task instead of a dynamic table.

• If you need more support, you can email us at snowconvert-support@snowflake.com

Description¶

Snowflake does not support the MERGE ... REMOVE DUPLICATES clause. SnowConvert AI generates a workaround that includes an INSERT WHEN NOT MATCHED clause, which requires knowledge of the source table’s columns. If the source table definition was not included in the code provided to SnowConvert AI, the column list cannot be generated and must be added manually.

Code Example¶

 MERGE INTO target USING source ON target.id = source.id REMOVE DUPLICATES;

 CREATE TEMPORARY TABLE source_duplicates AS
SELECT DISTINCT
source.*
FROM
source
INNER JOIN
target
ON target.id = source.id;
!!!RESOLVE EWI!!! /*** SSC-EWI-RS0009 - SEMANTIC INFORMATION NOT FOUND FOR THE SOURCE TABLE IN THE CODE PROVIDED TO SNOWCONVERT AI. COLUMNS TO BE INSERTED MAY BE ADDED MANUALLY. ***/!!!
--** SSC-FDM-RS0005 - REDSHIFT MERGE STATEMENT REJECTS DUPLICATE SOURCE ROWS. SNOWFLAKE ALLOWS DUPLICATES, WHICH MAY PRODUCE NON-DETERMINISTIC RESULTS. **
MERGE INTO target
USING source ON target.id = source.id
WHEN MATCHED THEN
DELETE
WHEN NOT MATCHED THEN
INSERT
VALUES ();
INSERT INTO target
SELECT
*
FROM
source_duplicates;
DROP TABLE IF EXISTS source_duplicates CASCADE;

Best Practices¶

• Include all source DDL: Provide the source table’s CREATE TABLE statement in the input code so SnowConvert AI can resolve columns automatically.

• Add columns manually: If the source table definition is unavailable, fill in the INSERT ... VALUES () clause with the correct column list from your Redshift catalog.

• If you need more support, you can email us at snowconvert-support@snowflake.com

Description¶

Redshift allows top-level CALL statements to invoke procedures with OUT parameters without declaring a variable to receive the output. Snowflake requires that OUT parameters be assigned to a variable, which is only possible inside a stored procedure or anonymous block.

Code Example¶

 CREATE OR REPLACE PROCEDURE get_total_sales_by_product(
    IN p_product_name VARCHAR(100),
    OUT p_total_sales DECIMAL(18, 2)
)
AS $$
BEGIN
    NULL;
END;
$$ LANGUAGE plpgsql;

CALL get_total_sales_by_product('Laptop');

 CREATE OR REPLACE PROCEDURE get_total_sales_by_product (p_product_name VARCHAR(100), p_total_sales OUT DECIMAL(18, 2))
RETURNS VARCHAR
LANGUAGE SQL
COMMENT = '{ "origin": "sf_sc", "name": "snowconvert", "version": {  "major": 0,  "minor": 0,  "patch": "0" }, "attributes": {  "component": "redshift",  "convertedOn": "07/10/2025",  "domain": "no-domain-provided" }}'
AS $$
BEGIN
NULL;
END;
$$;
!!!RESOLVE EWI!!! /*** SSC-EWI-RS0010 - TOP-LEVEL PROCEDURE CALL WITH OUT PARAMETERS IS NOT SUPPORTED IN SNOWFLAKE. ***/!!!
CALL get_total_sales_by_product('Laptop');

Best Practices¶

• Move the call into an anonymous block and declare a variable to pass as an output parameter.

• If you need more support, you can email us at snowconvert-support@snowflake.com