fix(sink): derive append only from optimizer when no format declared

[st1page]

I hereby agree to the terms of the RisingWave Labs, Inc. Contributor License Agreement.

What's changed and what's your intention?

• Add a branch for the case that the user does not declare sink type/sink format,
• allow force append only when there is no FORMAT clause

which is especially useful for "CREATE SINK INTO TABLE"

create table t(a int) append only;
create table tt(a int) append only;
create table ttt(a int);

dev=> 
explain create sink s into t as select a from tt;
                             QUERY PLAN                              
---------------------------------------------------------------------
 StreamProject { exprs: [tt.a, null:Serial] }
 └─StreamSink { type: AppendOnly, columns: [a, tt._row_id(hidden)] }
   └─StreamTableScan { table: tt, columns: [a, _row_id] }
(3 rows)

dev=> 
explain create sink s into t as select a from ttt with (force_append_only = true);
                                QUERY PLAN                                 
---------------------------------------------------------------------------
 StreamProject { exprs: [ttt.a, null:Serial] }
 └─StreamSink { type: ForceAppendOnly, columns: [a, ttt._row_id(hidden)] }
   └─StreamTableScan { table: ttt, columns: [a, _row_id] }
(3 rows)

Checklist

• I have written necessary rustdoc comments
• I have added necessary unit tests and integration tests
• I have added test labels as necessary. See details.
• I have added fuzzing tests or opened an issue to track them. (Optional, recommended for new SQL features Sqlsmith: Sql feature generation #7934).
• My PR contains breaking changes. (If it deprecates some features, please create a tracking issue to remove them in the future).
• All checks passed in ./risedev check (or alias, ./risedev c)
• My PR changes performance-critical code. (Please run macro/micro-benchmarks and show the results.)

• My PR contains critical fixes that are necessary to be merged into the latest release. (Please check out the details)

Documentation

• My PR needs documentation updates. (Please use the Release note section below to summarize the impact on users)

Release note

If this PR includes changes that directly affect users or other significant modifications relevant to the community, kindly draft a release note to provide a concise summary of these changes. Please prioritize highlighting the impact these changes will have on users.

[codecov]

Codecov Report

Attention: 12 lines in your changes are missing coverage. Please review.

Comparison is base (ef6d483) 68.10% compared to head (89922a6) 68.10%.
Report is 30 commits behind head on main.

Files	Patch %	Lines
...rc/frontend/src/optimizer/plan_node/stream_sink.rs	69.23%	12 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #14065      +/-   ##
==========================================
- Coverage   68.10%   68.10%   -0.01%     
==========================================
  Files        1543     1543              
  Lines      267107   267111       +4     
==========================================
+ Hits       181915   181917       +2     
- Misses      85192    85194       +2     

Flag	Coverage Δ
rust	68.10% <69.23%> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[xiangjinwu]

Code itself lgtm as long as all the behavior changes are intentional. Please confirm before merge.

[xiangjinwu]

Strictly speaking, this is a backward incompatible change: NotDefined was considered false before this PR, even when frontend derives it is actually append only. It used to return SinkType::Upsert in this case. This may or may not be a real problem.

One such example:
https://github.com/risingwavelabs/risingwave/blob/main/integration_tests/elasticsearch-sink/create_sink.sql

To list all behavior changes:

frontend derived	user defined	user force	old	new
true	undefined	true	err cannot force	SinkType::AppendOnly
true	undefined	false	SinkType::Upsert	SinkType::AppendOnly
false	undefined	true	err cannot force	SinkType::ForceAppendOnly
false	undefined	false	SinkType::Upsert	SinkType::Upsert (no change)

Based on our previous discussion, we wanted to make format required if backward compatibility was not a concern. This PR is actually making the backward compatibility promise harder to maintain, as undefined can now mean AppendOnly/ForceAppendOnly/Upsert in different cases, which used to be always Upsert. cc @tabVersion

[st1page]

Thanks for your clarification. If we do not maintain the compatibility for those errored statements, the only change is this one.

this is a backward incompatible change: NotDefined was considered false before this PR, even when frontend derives it is actually append only.

[xiangjinwu]

If we do not maintain the compatibility for those errored statements

In my opinion, it is okay now to move from err to non-err in this PR. However, if we want to move from non-err to err again in the future, it would be a breaking change. This is what I mean by making the backward compatibility promise harder to maintain.

[st1page]

After offline, it will be implemented with a less invasive method