steFaiz opened a new pull request, #8354:
URL: https://github.com/apache/paimon/pull/8354
Reverts apache/paimon#7221
I believe the scenario fixed by that PR is not reachable under the current
lookup compaction layout, and the added logic makes the lookup merge semantics
more complex than necessary.
The original PR is trying to fix this scenario:
```text
L1: sequence = 7
L2: sequence = 8
L0: sequence = 6
```
However, in lookup mode, high-level files are expected to represent
materialized merged states. A lower high level, such as L1, should shadow older
higher levels, such as L2. If an L1 record exists for a key after compaction
involving L2, it should already have been produced by merging the relevant L0
and higher-level state, and therefore should carry the latest state according
to the configured merge function and UserDefinedSeqComparator (the max
seqNumber).
I've checked this PR's tests, the PR manually constructs records like
```text
L1: sequence = 7
L2: sequence = 8
L0: sequence = 6
```
and feeds them directly into LookupChangelogMergeFunctionWrapper.
These tests bypass the actual writer, level manager, compaction picker,
rewrite/upgrade strategy, and lookup compaction flow.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
