sql: add optional message to CANCEL commands

[XiaochenCui]

overview

Previously, asynchronous CANCEL commands ("CANCEL QUERY", "CANCEL SESSION") didn't provide much information back to the query/session being canceled.

This commit addresses this issue by:

• Update the SQL syntax of "CANCEL" commands ("CANCEL QUERY", "CANCEL SESSION") to add a new "WITH MESSAGE " clause.
• The canceled query/session will throw an error with the from the canceler.

Fixes: #127490

Release note (sql change): Add optional "WITH MESSAGE " clause to CANCEL commands, providing more detailed information when a query is canceled.

sql syntax changes

origin:
// CANCEL QUERIES [IF EXISTS] <selectclause>
// CANCEL QUERY [IF EXISTS] <expr>

new:
// CANCEL QUERIES [IF EXISTS] <selectclause> [WITH MESSAGE <cancel_message>]
// CANCEL QUERY [IF EXISTS] <expr> [WITH MESSAGE <cancel_message>]

how to review

1. I updated the sql syntax: pkg/sql/parser/sql.y
2. A new logic test is added to test this feature: pkg/sql/logictest/testdata/logic_test/cancel
3. The "cancel message" is sent from the canceler: pkg/sql/conn_executor.go
4. The "target query" receives the "cancel message" and throw it to client: pkg/sql/conn_executor_exec.go

Other changes are trivial.

concerns

• This pr involves sql syntax changes, should we open a new pr and work on a RFC at first? (see: 20171011_adding_sql_syntax.md ), that's also the reason I post this unfinished commit for review

  one that focuses on changes to the SQL grammar; in short, new syntactic features should be discussed via the RFC process and merged with the same PR that actually implements the feature.

• The current implementation uses a synchronized map to send the "cancel message" to the query being canceled. This might be too heavy for such a small feature, and further experimentation is needed to assess the performance impact. (or we can just get the work done, depend on the development style)

todo

• Design new syntax for "CANCEL QUERY" to support adding custom messages.
• Design new syntax for "CANCEL SESSION" to support adding custom messages.

[blathers-crl]

Thank you for contributing to CockroachDB. Please ensure you have followed the guidelines for creating a PR.

My owl senses detect your PR is good for review. Please keep an eye out for any test failures in CI.

I have added a few people who may be able to assist in reviewing:

• @michae2 (author of sql: add optional message to CANCEL commands #127490)

_{🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is dev-inf.}

[cockroach-teamcity]

This change is 

[blathers-crl]

Thank you for updating your pull request.

My owl senses detect your PR is good for review. Please keep an eye out for any test failures in CI.

_{🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is dev-inf.}

[blathers-crl]

It looks like your PR touches SQL parser code but doesn't add or edit parser tests. Please make sure you add or edit parser tests if you edit the parser.

Thank you for updating your pull request.

My owl senses detect your PR is good for review. Please keep an eye out for any test failures in CI.

_{🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is dev-inf.}

[XiaochenCui]

concerns

• This pr involves sql syntax changes, should we open a new pr and work on a RFC at first? (see: 20171011_adding_sql_syntax.md ), that's also the reason I post this unfinished commit for review

  one that focuses on changes to the SQL grammar; in short, new syntactic features should be discussed via the RFC process and merged with the same PR that actually implements the feature.

• The current implementation uses a synchronized map to send the "cancel message" to the query being canceled. This might be too heavy for such a small feature, and further experimentation is needed to assess the performance impact. (or we can just get the work done, depend on the development style)

hi, @michae2
I've updated the sql syntax and now we can pass message to the query being canceled. All lints and test have also be passed.