Support Sort pushdown #3620

yuancu · 2025-05-13T07:22:58Z

Description

Support Sort pushdown

Related Issues

Resolves #3380

Check List

New functionality includes testing.
New functionality has been documented.
New functionality has javadoc added.
New functionality has a user manual doc added.
API changes companion pull request created.
Commits are signed per the DCO using --signoff.
Public documentation issue/PR created.

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
For more information on following Developer Certificate of Origin and signing off your commits, please check here.

Signed-off-by: Yuanchun Shen <[email protected]> Copy traits to logical index scan after pushing down sort Signed-off-by: Yuanchun Shen <[email protected]> Allow pushing down multiple times Signed-off-by: Yuanchun Shen <[email protected]>

Signed-off-by: Yuanchun Shen <[email protected]>

Additionally correct merge collation method Signed-off-by: Yuanchun Shen <[email protected]>

Signed-off-by: Yuanchun Shen <[email protected]>

… in sort by fields Signed-off-by: Yuanchun Shen <[email protected]>

Signed-off-by: Yuanchun Shen <[email protected]>

penghuo · 2025-06-23T21:59:39Z

...search/src/main/java/org/opensearch/sql/opensearch/storage/scan/CalciteLogicalIndexScan.java

+    // `Sort($1)\n TableScan(name, age)` may become
+    // `Sort($0)\n Project(age)\n  TableScan(name, age)` after projection.


Why Sort($1)\n TableScan(name, age) may become Sort($0)\n Project(age)\n TableScan(name, age)? SortRule should pushdown sort already?

During optimization, the index in collations should point to its location in the current row instead of its location in the original table. As a result, even if sort is already pushed down, the trait RelCollation, which represents the sort status of the relation being optimized, should be updated to reflect the latest order status when there is a projection after sort.

...search/src/main/java/org/opensearch/sql/opensearch/storage/scan/CalciteLogicalIndexScan.java

penghuo · 2025-06-23T22:43:31Z

...search/src/main/java/org/opensearch/sql/opensearch/storage/scan/CalciteLogicalIndexScan.java

+      if (LOG.isDebugEnabled()) {
+        LOG.debug("Cannot pushdown the sort {}", collations, e);
+      } else {
+        LOG.warn("Cannot pushdown the sort {}, ", collations);
+      }


santize collations? does it include field name or value?

Fixed. Replaced with their field names.

penghuo · 2025-06-23T22:47:00Z

...search/src/main/java/org/opensearch/sql/opensearch/storage/scan/CalciteLogicalIndexScan.java

+                ? SortOrder.DESC
+                : SortOrder.ASC;
+        // TODO: support script sort and distance sort
+        SortBuilder<?> sortBuilder;


is there IT/UT cover this?

No, I haven't found any. Do you know the proper syntax for sorting geopoints or with script? I tried sorting on geopoints, it reports can't sort on geo_point field without using specific sorting feature, like geo_distance. But I haven't figured out the correct syntax for sorting geopoints (it's not in the g4 file or the documentation).

penghuo · 2025-06-23T22:48:34Z

integ-test/src/test/resources/expectedOutput/calcite/explain_sort_push.json

@@ -1,6 +1,6 @@
 {
  "calcite": {
    "logical": "LogicalSort(sort0=[$0], dir0=[ASC])\n  LogicalProject(age=[$8])\n    LogicalFilter(condition=[>($8, 30)])\n      LogicalSort(sort0=[$8], dir0=[ASC])\n        CalciteLogicalIndexScan(table=[[OpenSearch, opensearch-sql_test_index_account]])\n",
-    "physical": "EnumerableCalc(expr#0=[{inputs}], expr#1=[30], expr#2=[>($t0, $t1)], age=[$t0], $condition=[$t2])\n  EnumerableSort(sort0=[$0], dir0=[ASC])\n    CalciteEnumerableIndexScan(table=[[OpenSearch, opensearch-sql_test_index_account]], PushDownContext=[[PROJECT->[age]], OpenSearchRequestBuilder(sourceBuilder={\"from\":0,\"timeout\":\"1m\",\"_source\":{\"includes\":[\"age\"],\"excludes\":[]}}, requestedTotalSize=2147483647, pageSize=null, startFrom=0)])\n"
+    "physical": "CalciteEnumerableIndexScan(table=[[OpenSearch, opensearch-sql_test_index_account]], PushDownContext=[[PROJECT->[age], SORT->[{\n  \"age\" : {\n    \"order\" : \"asc\"\n  }\n}], FILTER->>($0, 30)], OpenSearchRequestBuilder(sourceBuilder={\"from\":0,\"timeout\":\"1m\",\"query\":{\"range\":{\"age\":{\"from\":30,\"to\":null,\"include_lower\":false,\"include_upper\":true,\"boost\":1.0}}},\"_source\":{\"includes\":[\"age\"],\"excludes\":[]},\"sort\":[{\"age\":{\"order\":\"asc\"}}]}, requestedTotalSize=2147483647, pageSize=null, startFrom=0)])\n"


Add another IT on timestamp field sort? source=xxx | sort @timestamp

qianheng-aws · 2025-06-24T03:26:26Z

...search/src/main/java/org/opensearch/sql/opensearch/storage/scan/CalciteLogicalIndexScan.java

+      CalciteLogicalIndexScan newScan =
+          new CalciteLogicalIndexScan(
+              getCluster(),
+              traitsWithCollations,


After adding collation by pushing down sort, we should remove them if pushing down aggregation afterwards, as aggregation will eliminate any collation.

Additionally remove sort when pushing down aggregation after adding collation by pushing down sort Signed-off-by: Yuanchun Shen <[email protected]>

…into pushdown-sort

Signed-off-by: Yuanchun Shen <[email protected]>

yuancu · 2025-06-25T03:21:56Z

I found a corner case bug when sorting nested fields: sorting nested text fields does not leverage its keyword.

For example (mapping):
For query: source=opensearch-sql_test_index_nested_simple | sort address.state, its DSL is as follows

{
  "from": 0,
  "timeout": "1m",
  "_source": {
    "includes": [
      "name",
      "address",
      "id",
      "age"
    ],
    "excludes": []
  },
  "sort": [
    {
      "address.state": {
        "order": "asc"
      }
    }
  ]
}

But when executing the plan, I get the following response:

{
  "error": {
    "reason": "Error occurred in OpenSearch engine: all shards failed",
    "details": "Shard[0]: java.lang.IllegalArgumentException: Text fields are not optimised for operations that require per-document field data like aggregations and sorting, so these operations are disabled by default. Please use a keyword field instead. Alternatively, set fielddata=true on [address.state] in order to load field data by uninverting the inverted index. Note that this can use significant memory.\n\nFor more details, please send request for Json format to see the raw response from OpenSearch engine.",
    "type": "SearchPhaseExecutionException"
  },
  "status": 400
}

In contrast, the DSL in v2 uses the keyword field instead of the text field itself.

{
  "from": 0,
  "size": 10000,
  "timeout": "1m",
  "_source": {
    "includes": [
      "name",
      "address",
      "id",
      "age"
    ],
    "excludes": []
  },
  "sort": [
    {
      "address.state.keyword": {
        "order": "asc",
        "missing": "_first"
      }
    }
  ]
}

The plan in v2 can be executed with the following result. Although it's not quite correct. The engine seems to pick a random address.state from the nested list and displays another one when project with fields.

{
    "schema": [
        {
            "name": "address.state",
            "type": "string"
        }
    ],
    "datarows": [
        [
            "Florida"
        ],
        [
            "CA"
        ],
        [
            "NY"
        ],
        [
            "NC"
        ],
        [
            "TX"
        ]
    ],
    "total": 5,
    "size": 5
}

Update
Fixed by pointing to its keyword field when sorting a text field (if possible).

Signed-off-by: Yuanchun Shen <[email protected]>

yuancu · 2025-06-25T08:40:32Z

Another corner case to be fixed: sort is not pushed down when when there is a filter before it.

The physical plan for the query source=bank | where birthdate < '2023-01-01 23:59:58.000000000' | sort birthdate | fields birthdate is:

EnumerableSort(sort0=[$0], dir0=[ASC])
  EnumerableCalc(expr#0=[{inputs}], expr#1=[TIMESTAMP($t0)], expr#2=['2023-01-01 23:59:58':EXPR_TIMESTAMP VARCHAR], expr#3=[<($t1, $t2)], birthdate=[$t0], $condition=[$t3])
    CalciteEnumerableIndexScan(table=[[OpenSearch, bank]], PushDownContext=[[PROJECT->[birthdate]], OpenSearchRequestBuilder(sourceBuilder={"from":0,"timeout":"1m","_source":{"includes":["birthdate"],"excludes":[]}}, requestedTotalSize=2147483647, pageSize=null, startFrom=0)])

However, in v2, the plan for it is :

{
    "root": {
        "name": "ProjectOperator",
        "description": {
            "fields": "[birthdate]"
        },
        "children": [
            {
                "name": "OpenSearchIndexScan",
                "description": {
                    "request": "OpenSearchQueryRequest(indexName=bank, sourceBuilder={\"from\":0,\"size\":10000,\"timeout\":\"1m\",\"query\":{\"range\":{\"birthdate\":{\"from\":null,\"to\":1672617598000,\"include_lower\":true,\"include_upper\":false,\"boost\":1.0}}},\"_source\":{\"includes\":[\"birthdate\"],\"excludes\":[]},\"sort\":[{\"birthdate\":{\"order\":\"asc\",\"missing\":\"_first\"}}]}, needClean=true, searchDone=false, pitId=*, cursorKeepAlive=1m, searchAfter=null, searchResponse=null)"
                },
                "children": []
            }
        ]
    }
}

From the plan, where ... | sort ... is pushed down in v2 but not in Calcite.

Update

The reason turns out to be related to timestamp comparison. where account_number > 10 | sort account_number can be properly pushed down.
Will be fixed by Change compare logical when comparing date related fields with string literal #3798

LantaoJin · 2025-06-27T02:38:59Z

integ-test/src/test/resources/expectedOutput/calcite/explain_sort_push.json

@@ -1,6 +1,6 @@
 {
  "calcite": {
    "logical": "LogicalSort(sort0=[$0], dir0=[ASC])\n  LogicalProject(age=[$8])\n    LogicalFilter(condition=[>($8, 30)])\n      LogicalSort(sort0=[$8], dir0=[ASC])\n        CalciteLogicalIndexScan(table=[[OpenSearch, opensearch-sql_test_index_account]])\n",
-    "physical": "EnumerableCalc(expr#0=[{inputs}], expr#1=[30], expr#2=[>($t0, $t1)], age=[$t0], $condition=[$t2])\n  EnumerableSort(sort0=[$0], dir0=[ASC])\n    CalciteEnumerableIndexScan(table=[[OpenSearch, opensearch-sql_test_index_account]], PushDownContext=[[PROJECT->[age]], OpenSearchRequestBuilder(sourceBuilder={\"from\":0,\"timeout\":\"1m\",\"_source\":{\"includes\":[\"age\"],\"excludes\":[]}}, requestedTotalSize=2147483647, pageSize=null, startFrom=0)])\n"
+    "physical": "CalciteEnumerableIndexScan(table=[[OpenSearch, opensearch-sql_test_index_account]], PushDownContext=[[PROJECT->[age], SORT->[{\n  \"age\" : {\n    \"order\" : \"asc\",\n    \"missing\" : \"_last\"\n  }\n}], FILTER->>($0, 30)], OpenSearchRequestBuilder(sourceBuilder={\"from\":0,\"timeout\":\"1m\",\"query\":{\"range\":{\"age\":{\"from\":30,\"to\":null,\"include_lower\":false,\"include_upper\":true,\"boost\":1.0}}},\"_source\":{\"includes\":[\"age\"],\"excludes\":[]},\"sort\":[{\"age\":{\"order\":\"asc\",\"missing\":\"_last\"}}]}, requestedTotalSize=2147483647, pageSize=null, startFrom=0)])\n"


\"sort\":[{\"age\":{\"order\":\"asc\",\"missing\":\"_last\"}}]

Here _last is different with v2's behaviour, in v2, default null direction (missing) for ASC is first. But in Calcite, the ASC maps to last:
https://github.com/apache/calcite/blob/2fb437b93774d3d69df66547715b374d9281fdbb/core/src/main/java/org/apache/calcite/rel/RelFieldCollation.java#L134-L136

We can add a TODO to align with v2, but my question is that is the behaviour of v2 correct? @dai-chen @penghuo any thoughts about it ?

this issue is not a blocker for this PR.

LantaoJin · 2025-06-27T02:45:15Z

integ-test/src/test/resources/expectedOutput/ppl/explain_sort_rename_push.json

+                  {
+                    "name": "OpenSearchIndexScan",
+                    "description": {
+                      "request": "OpenSearchQueryRequest(indexName=opensearch-sql_test_index_account, sourceBuilder={\"from\":0,\"size\":10000,\"timeout\":\"1m\"}, needClean=true, searchDone=false, pitId=s9y3QQEhb3BlbnNlYXJjaC1zcWxfdGVzdF9pbmRleF9hY2NvdW50Fndla1VpMi1kVHh5Qi1lYnhPQnlSbXcAFkU4Qm9UVURIUWI2a3pjNkhmQkxvc2cAAAAAAAAAAAsWcnUyVHZfNk1SeC1DMFNFdGtPcDN4QQEWd2VrVWkyLWRUeHlCLWVieE9CeVJtdwAA, cursorKeepAlive=1m, searchAfter=null, searchResponse=null)"


Good catching, the v2 can not pushdown with rename.

LantaoJin · 2025-06-27T02:48:48Z

integ-test/src/test/java/org/opensearch/sql/ppl/ExplainIT.java

+    // TODO: Remove pushed-down sort in DSL in expectedOutput/ppl/explain_sort_then_agg_push.json
+    //  existing collations should be eliminated when pushing down aggregations


is this TODO for v2?

Yes. Added a (v2) in comment.

LantaoJin · 2025-06-27T02:55:18Z

integ-test/src/test/java/org/opensearch/sql/ppl/ExplainIT.java

+    // TODO: Fix the expected output in expectedOutput/ppl/explain_multi_sort_push.json
+    //  balance and gender should take precedence over account_number and firstname


Seems another bug of v2 sort.

source=opensearch_dashboards_sample_data_logs | head 10 | sort bytes | sort - bytes

with DSL pushdown

{ "from": 0, "size": 10, "timeout": "1m", "sort": [{ "bytes": { "order": "asc", "missing": "_first" } }, { "bytes": { "order": "desc", "missing": "_last" } }] }

the final result is sorted ASC.

Yes, in v2 it merges collations by simply concatenating them without de-duplication and re-order.

LantaoJin · 2025-06-27T03:01:50Z

...search/src/main/java/org/opensearch/sql/opensearch/storage/scan/CalciteLogicalIndexScan.java

+                case FIRST -> "_first";
+                case LAST -> "_last";


a workaround to align with v2 is changing to

case FIRST -> "_last"; case LAST -> "_first";

But not recommended.

LantaoJin · 2025-06-27T03:02:58Z

...search/src/main/java/org/opensearch/sql/opensearch/storage/scan/CalciteLogicalIndexScan.java

+      if (LOG.isDebugEnabled()) {
+        LOG.debug("Cannot pushdown the sort {}", getCollationNames(collations), e);
+      } else {
+        LOG.warn("Cannot pushdown the sort {}, ", getCollationNames(collations));


change to info.

LantaoJin · 2025-06-27T03:06:31Z

...search/src/main/java/org/opensearch/sql/opensearch/storage/scan/CalciteLogicalIndexScan.java

+
+  /**
+   * Check if the sort by collations contains any aggregators that are pushed down. E.g. In `stats
+   * avg(age) as avg_age by state | sort avg_age`, the sort clause has `avg_age` which is an


it may have a more complex case:

avg(age) as avg_age by state | rename avg_age as new_avg_age | sort new_avg_age

I don't think the current hasAggregatorInSortBy works for that.

It works. This case is covered by testSortWithAggregationExplain. The query source=opensearch-sql_test_index_account| stats avg(age) AS avg_age by state, city | sort avg_age won't be pushed down:

EnumerableSort(sort0=[$0], dir0=[ASC]) EnumerableCalc(expr#0..2=[{inputs}], avg_age=[$t2], state=[$t0], city=[$t1]) CalciteEnumerableIndexScan(table=[[OpenSearch, opensearch-sql_test_index_account]], PushDownContext=[[PROJECT->[state, city, age], AGGREGATION->rel#1064:LogicalAggregate.NONE.[](input=RelSubset#1063,group={0, 1},avg_age=AVG($2))], OpenSearchRequestBuilder(sourceBuilder={"from":0,"size":0,"timeout":"1m","_source":{"includes":["state","city","age"],"excludes":[]},"aggregations":{"composite_buckets":{"composite":{"size":1000,"sources":[{"state":{"terms":{"field":"state.keyword","missing_bucket":true,"missing_order":"first","order":"asc"}}},{"city":{"terms":{"field":"city.keyword","missing_bucket":true,"missing_order":"first","order":"asc"}}}]},"aggregations":{"avg_age":{"avg":{"field":"age"}}}}}}, requestedTotalSize=2147483647, pageSize=null, startFrom=0)])

When sorting avg_age, the field names by getRowType().getFieldNames() is already the renamed ones (state, city, avg_age).

LantaoJin · 2025-06-27T03:14:27Z

@yuancu do we have follow cases in ExplainIT

source=t | head 10 | sort field // pushdown limit only
source=t | sort field | head 10 // pushdown sort only or pushdown sort + limit (depends on the execution order of DSL)

…sort Signed-off-by: Yuanchun Shen <[email protected]>

yuancu · 2025-06-27T08:06:58Z

@yuancu do we have follow cases in ExplainIT

source=t | head 10 | sort field // pushdown limit only
source=t | sort field | head 10 // pushdown sort only or pushdown sort + limit (depends on the execution order of DSL)

Added the test cases & fixed the behavior.

…e IT Signed-off-by: Yuanchun Shen <[email protected]>

LantaoJin added the calcite calcite migration releated label May 22, 2025

yuancu force-pushed the pushdown-sort branch from 83cf083 to a287caf Compare May 28, 2025 08:28

yuancu force-pushed the pushdown-sort branch from a287caf to c4ae5ca Compare June 9, 2025 10:02

yuancu added 3 commits June 20, 2025 14:58

Support Sort pushdown

fd620b4

Signed-off-by: Yuanchun Shen <[email protected]> Copy traits to logical index scan after pushing down sort Signed-off-by: Yuanchun Shen <[email protected]> Allow pushing down multiple times Signed-off-by: Yuanchun Shen <[email protected]>

Merge multiple sort when occurs

8553e0a

Signed-off-by: Yuanchun Shen <[email protected]>

Update trendline explain IT

4c7f759

Additionally correct merge collation method Signed-off-by: Yuanchun Shen <[email protected]>

yuancu force-pushed the pushdown-sort branch from c4ae5ca to 4c7f759 Compare June 20, 2025 06:59

yuancu added 2 commits June 23, 2025 14:31

Re-index collations when pushing down projection

0d46563

Signed-off-by: Yuanchun Shen <[email protected]>

Support sort pushdown with aggregation when aggregated fields are not…

70b17ca

… in sort by fields Signed-off-by: Yuanchun Shen <[email protected]>

yuancu changed the title ~~WIP: Support Sort pushdown~~ Support Sort pushdown Jun 23, 2025

yuancu marked this pull request as ready for review June 23, 2025 14:59

yuancu requested review from ps48, kavithacm, derek-ho, joshuali925, dai-chen, YANG-DB, mengweieric, Swiddis, penghuo, seankao-az, MaxKsyunz, Yury-Fridlyand, anirudha, forestmvey, acarbonetto, GumpacG, ykmr1224, LantaoJin and noCharger as code owners June 23, 2025 14:59

yuancu requested a review from qianheng-aws as a code owner June 23, 2025 14:59

Update explain_output.json

4473718

Signed-off-by: Yuanchun Shen <[email protected]>

penghuo reviewed Jun 23, 2025

View reviewed changes

qianheng-aws reviewed Jun 24, 2025

View reviewed changes

yuancu marked this pull request as draft June 24, 2025 05:11

yuancu added 6 commits June 24, 2025 17:47

Support multi-sort

2daabec

Additionally remove sort when pushing down aggregation after adding collation by pushing down sort Signed-off-by: Yuanchun Shen <[email protected]>

Merge branch 'pushdown-sort' of https://github.com/yuancu/sql-plugin …

1ba081f

…into pushdown-sort

Sanitize collations when sort can't be pushed down

8167f97

Signed-off-by: Yuanchun Shen <[email protected]>

Test explaining sort | stats avg

e13dc72

Signed-off-by: Yuanchun Shen <[email protected]>

Test sort date

764e450

Signed-off-by: Yuanchun Shen <[email protected]>

Chores: reorder util functions

f91b2cf

Signed-off-by: Yuanchun Shen <[email protected]>

yuancu marked this pull request as ready for review June 24, 2025 13:39

penghuo added the backport 2.19-dev label Jun 24, 2025

yuancu added 3 commits June 25, 2025 12:25

Pass null direction when pushing down sort

e080439

Signed-off-by: Yuanchun Shen <[email protected]>

Convert to keyword fields when sorting opensearch text fields

89d487b

Signed-off-by: Yuanchun Shen <[email protected]>

Test explaining sort after rename

bc07aee

Signed-off-by: Yuanchun Shen <[email protected]>

yuancu mentioned this pull request Jun 25, 2025

[FEATURE] Calcite Engine Framework: pushdown sort #3380

Open

penghuo previously approved these changes Jun 26, 2025

View reviewed changes

Swiddis added the enhancement New feature or request label Jun 26, 2025

LantaoJin reviewed Jun 27, 2025

View reviewed changes

Push down sort-then-limit, while only pushdown limit when limit-then-…

8ad01f7

…sort Signed-off-by: Yuanchun Shen <[email protected]>

yuancu dismissed penghuo’s stale review via 8ad01f7 June 27, 2025 07:56

yuancu added 2 commits June 27, 2025 17:18

Merge remote-tracking branch 'origin/main' into pushdown-sort

775d504

Fix flaky test related to irregular number in some locales in datetim…

9552497

…e IT Signed-off-by: Yuanchun Shen <[email protected]>

		// `Sort($1)\n TableScan(name, age)` may become
		// `Sort($0)\n Project(age)\n TableScan(name, age)` after projection.

		// TODO: Remove pushed-down sort in DSL in expectedOutput/ppl/explain_sort_then_agg_push.json
		// existing collations should be eliminated when pushing down aggregations

		// TODO: Fix the expected output in expectedOutput/ppl/explain_multi_sort_push.json
		// balance and gender should take precedence over account_number and firstname

Support Sort pushdown #3620

Are you sure you want to change the base?

Support Sort pushdown #3620

Conversation

yuancu commented May 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Related Issues

Check List

Uh oh!

Choose a reason for hiding this comment

Uh oh!

yuancu Jun 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

yuancu Jun 24, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

yuancu commented Jun 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

yuancu commented Jun 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

LantaoJin Jun 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

LantaoJin commented Jun 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

yuancu commented Jun 27, 2025

Uh oh!

Uh oh!

yuancu commented May 13, 2025 •

edited

Loading

yuancu Jun 24, 2025 •

edited

Loading

yuancu Jun 24, 2025 •

edited

Loading

yuancu commented Jun 25, 2025 •

edited

Loading

yuancu commented Jun 25, 2025 •

edited

Loading

LantaoJin Jun 27, 2025 •

edited

Loading

LantaoJin commented Jun 27, 2025 •

edited

Loading