feat: eliminate adjacent no-op updates for agg with empty agg call

[BugenZhao]

Signed-off-by: Bugen Zhao [email protected]I hereby agree to the terms of the RisingWave Labs, Inc. Contributor License Agreement.

What's changed and what's your intention?

See background in #17030. Close #17030.

Introduced a flag named noop_update_hint on StreamProject indicating...

Whether there are likely no-op updates in the output chunks, so that eliminating them with StreamChunk::eliminate_adjacent_noop_update could be beneficial.

and only enable it on the one following HashAgg with empty agg call set, which usually happens in queries with semantics of deduplication.

This reduces the need for always calling eliminate on the output, resulting in minimal performance overhead.

Checklist

• I have written necessary rustdoc comments
• I have added necessary unit tests and integration tests
• All checks passed in ./risedev check (or alias, ./risedev c)

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.

[chenzl25]

LGTM, interesting insight~

[stdrc]

LGTM!

[fuyufjh]

I can understand the motivation, but why it's done in ProjectExecutor? In my mind Project means evaluating expression - 1 row in, 1 row out. This PR seems to add some irrelevant logic to Project.

[stdrc]

I can understand the motivation, but why it's done in ProjectExecutor? In my mind Project means evaluating expression - 1 row in, 1 row out. This PR seems to add some irrelevant logic to Project.

We can consider Projects as convenient hook points to do such no-op updates elimination, because they naturally pervade the whole query.😄