[CASSANDRA-16439] Reduce new reserved keywords introduced since 3.0

• Type: Improvement
• Status: Resolved
• Priority: Normal
• Resolution: Fixed
• Fix Version/s: 4.0-rc1, 4.0
• Component/s: CQL/Syntax
• Change Category: Operability
• Complexity: Normal
• Platform: All

Source Control Link

https://github.com/apache/cassandra/commit/b74d7370cc89fa899f47f50c825ddaed2dd05c3f

Test and Documentation Plan

Added KeywordTest that goes through the keyword tokens and checks creating, updating and selecting tables and materialized views with the keyword as the table/column names.

• Assignee: Jon Meredith
• Reporter: Jon Meredith

Authors

Jon Meredith

Reviewers

Alex Petrov, David Capwell, Yifan Cai

• Created: 11/Feb/21 00:23
• Updated: 25/Feb/21 10:10
• Resolved: 12/Feb/21 05:21

Description

Building on CASSANDRA-16398, for users upgrading from 3.0 to 4.0 there have been a few new reserved keywords added that can cause problems with schema changes and queries in existing applications. Reducing the number of new reserved words makes upgrades easier for users.

K_DEFAULT
K_MBEAN
K_MBEANS
K_NEGATIVE_INFINITY
K_NEGATIVE_NAN
K_POSITIVE_INFINITY
K_POSITIVE_NAN
K_REPLACE
K_SCHEMA
K_UNSET

I think they can all be removed from the ReservedWords lists (where present) and added to basic_unreserved_keyword with the exception of the NAN and INFINITY as there could be ambiguity between a selected field and a nan/infinity literal.

To validate I've written a test that extracts the keywords from the parser tokens and verifies that any non-reserved keywords can be used as table and column names for regular tables and materialized views. It isn't exhaustive but provides a good sense check.

Issue Links

• is duplicated by: CASSANDRA-16233 remove "replace" from cql reserved words