[indexer-alt] add coin jsonrpc impl #20931

emmazzz · 2025-01-21T08:39:14Z

Description

This PR added indexer alt jsonrpc reader implementation for getCoins and getAllCoins. I ported over raw_query from graphql crate for usage here.

Test plan

Added e2e tests.

Release notes

Check each box that your changes affect. If none of the boxes relate to your changes, release notes aren't required.

For each box you select, include information after the relevant heading that describes the impact of your changes that a user might notice and any actions they must take to implement updates.

vercel · 2025-01-21T08:39:18Z

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
sui-docs	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Feb 12, 2025 0:52am

2 Skipped Deployments

Name	Status	Preview	Comments	Updated (UTC)
multisig-toolkit	⬜️ Ignored (Inspect)	Visit Preview		Feb 12, 2025 0:52am
sui-kiosk	⬜️ Ignored (Inspect)	Visit Preview		Feb 12, 2025 0:52am

lxfind · 2025-01-21T21:06:08Z

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

+            // let cursor_coin_query = cb::table.select(StoredCoinBalanceBucket::as_select()).filter(cb::object_id.eq(cursor.to_vec()));
+            let mut cursor_coins: Vec<StoredCoinBalanceBucket> =
+                conn.results_with_raw_query(cursor_coin_query).await?;
+            if cursor_coins.len() != 1 {


Not sure what you are trying to achieve here, but there can and will be multiple entries with the same object_id in this table, since it is an append table, we add entries for new mutations.

I'm trying to get the balance bucket of the cursor coin here so that we can use the balance in the filter, because results are sorted by balance. And you are right I should account for the append-only-ness here. I've updated the query in the latest commit to select the row with the latest cp sequence number.

lxfind · 2025-01-21T21:06:33Z

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

+
+        // First load the coins' object ids.
+        // TODO: optimize this query according to the indexes in coin_balance_buckets table.
+        let mut query = query!("


I think you would have to rewrite the entire function with a full raw string as the query.

Updated in the latest commit.

crates/sui-indexer-alt-jsonrpc/src/reader.rs

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

crates/sui-indexer-alt-jsonrpc/src/reader.rs

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

emmazzz · 2025-02-03T07:01:28Z

Revamped and rebased this PR based on the latest changes in configs, pagination, errors and data loaders.
Another noticeable change is with the new cursor format consisting of (object_id, cp_sequence_number, balance_bucket) @gegaowp .

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

amnn

This is looking in pretty good shape, thanks @emmazzz. Here's a summary of the comments I left, from most to least important:

Query performance -- I think the implementation of getAllCoins may have an unnecessary sort in it that we can avoid, and I'm a little bit worried about the DB's ability to efficiently plan the comparison on bucket, checkpoint, ID (so worth double checking that).
- There's a tangent here about the size of the cursor and potentially using BCS encoded cursors which we can discuss.
Transactional test runner change -- this is a bit of a larger change and needs some more design -- I've proposed something in the inline comments. The issue with the current approach is that the feature being introduced (interpolating object IDs into cursors) bridges two places where we use similar syntax for different things, which is likely to cause confusion.
Errors -- I think are mostly there, but I made some suggestions to bring them fully in line (the biggest thing to address here is the conversion function not using unwraps, the second biggest thing is the behaviour when we fail to fetch contents which we need to align on).

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

crates/sui-indexer-alt-jsonrpc/src/data/objects.rs

crates/sui-indexer-alt-jsonrpc/src/lib.rs

crates/sui-indexer-alt-schema/src/objects.rs

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

crates/sui-indexer-alt-e2e-tests/tests/jsonrpc/coin/deleted.move

amnn · 2025-02-03T12:14:34Z

crates/sui-indexer-alt-e2e-tests/tests/jsonrpc/coin/get_all_coins.exp

+  "result": {
+    "data": [
+      {
+        "coinType": "0x0000000000000000000000000000000000000000000000000000000000000002::sui::SUI",


This output makes me think that we are not using the coin balance buckets index as efficiently as we could, because we are merging together different coin types, but IIRC coin type is the second field in the index after owner (so we should be returning all the results for one coin type in one go).

This also implies that the cursor should include the coin type. The cursor is also already quite large, so this may indicate that we need to introduce a BCS-encoded cursor (similar to what we did for GraphQL), but we can do that as a follow-up.

This test is for getAllCoins, which is different from getCoins in that it will return coins of all types in one go. getCoins on the other hand will only return coins of the same type each time so it shouldn't have any issues. getAllCoins is actually not used by Stashed. I only added this method because, well, it was convenient to. I'm happy to remove it or we can tackle this as a follow up later on after Stashed migration.

crates/sui-indexer-alt-e2e-tests/tests/jsonrpc/coin/inconsistency.move

crates/sui-indexer-alt-e2e-tests/tests/jsonrpc/coin/pagination.move

amnn

Recording our conversation from today!

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

amnn · 2025-02-05T18:35:20Z

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

+            query = query
+                .filter(candidates!(cp_sequence_number).le(cp_sequence_number as i64))
+                .filter(
+                    // Since the result is ordered by coin_balance_bucket followed by object_id,


The Filter on the first Index Only Scan means that we are performing a scan through the results that we may be able to avoid by utilising the index better (I think the piece we are missing for that is bounding by cp_sequence_number in the right order w.r.t. coin_balance_bucket and object_id -- i.e. in between).

crates/sui-indexer-alt-jsonrpc/src/data/objects.rs

crates/sui-indexer-alt-jsonrpc/src/data/reader.rs

crates/sui-indexer-alt-jsonrpc/src/lib.rs

crates/sui-indexer-alt-schema/src/objects.rs

amnn · 2025-02-05T19:45:10Z

crates/sui-transactional-test-runner/src/test_adapter.rs

@@ -1352,7 +1350,28 @@ impl<'a> SuiTestAdapter {

                variables.insert(format!("cursor_{idx}"), base64d);
            } else {
-                variables.insert(format!("cursor_{idx}"), Base64::encode(s));
+                // Handle cursor strings containing @{obj_x_y} pattern


Recording our conversation from the pair programming session: It's okay (beneficial, even) to only interpolate other variables when dealing with interpolation for cursors (i.e. pass in an empty list of cursors), and I think it's a good idea to factor this change out into its own commit, at least, but that commit should come first (because the test adapter is being actively touched by multiple people at the moment, so I would rather not increase the confusion factor there).

crates/sui-indexer-alt-jsonrpc/src/lib.rs

amnn · 2025-02-05T19:47:00Z

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

+            query = query
+                .filter(candidates!(cp_sequence_number).le(cp_sequence_number as i64))
+                .filter(
+                    // Since the result is ordered by coin_balance_bucket followed by object_id,


We talked about how we would probably need to invert the direction of object IDs in the index to make this as efficient as possible.

… cursors

emmazzz · 2025-02-10T06:01:11Z

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

+        })) = page.cursor
+        {
+            query = query.filter(sql::<Bool>(&format!(
+                r#"ROW(candidates.coin_balance_bucket, candidates.cp_sequence_number, candidates.object_id)


Changed to using ROW here and ordering by coin_balance_bucket, cp_sequence_number, object_id all DESC a few lines later as we've discussed @amnn

Rather than using &format!(...), use diesel's bind construct, so that we don't have to worry about re-implementing escaping/encoding. See below from #21119 for an example:

sui/crates/sui-indexer-alt-jsonrpc/src/api/objects/filter.rs

Lines 155 to 163 in bc3b4c6

if let Some(c) = page.cursor {

query = query.filter(

sql::<Bool>(r#"("candidates"."cp_sequence_number", "candidates"."object_id") < ("#)

.bind::<BigInt, _>(c.cp_sequence_number as i64)

.sql(", ")

.bind::<Bytea, _>(c.object_id.clone())

.sql(")"),

);

}

(I only learned about this recently, while working on this PR). It would look something like this for coin_balance_buckets:

if let Some(Cursor(c)) = page.cursor { query = query.filter( sql::<Bool>(r#"("candidates"."coin_balance_bucket", "candidates"."cp_sequence_number", "candidates"."object_id") < ("#) .bind::<BigInt, _>(c.coin_balance_bucket as i64) .sql(", ") .bind::<BigInt, _>(c.cp_sequence_number as i64) .sql(", ") .bind::<Bytea, _>(c.object_id.clone()) .sql(")"), ); }

(the ROW keyword is optional).

emmazzz · 2025-02-10T06:03:14Z

Updated the PR to address the second round of comments, include the index change on coin_balance_bucket and also rebased on top of the transactional test runner cursor pr #21150

amnn

Almost there -- main comments:

I suspect getAllCoins is still not going to perform well because of its ordering. We need to order by coin type first, then balance bucket, etc
The index changes need to be split into their own PR to unblock landing this one (also some suggestions for making the migration idempotent and easy to run proactively).
Various patterns/suggestions based on rpc-alt: getOwnedObjects #21119. Main ones are to use SqlLiteral::bind, represent the ObjectID using its DB form in the cursor, and to use context(...) instead of .ok_or_else(|| anyhow!(...)).

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

...indexer-alt-schema/migrations/2025-02-05-232254_coin_balance_buckets_object_id_desc/down.sql

crates/sui-indexer-alt-schema/src/objects.rs

amnn · 2025-02-11T13:15:41Z

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

+    }
+}
+
+impl Coins {


nit: it would be nice if we could structure this endpoint like the rest of them (with a response module to hold these functions). The rest of the methods are structured that way to promote sharing (it's not possible to share response functions when they are tied to the type implementing the method), and it would be good to maintain consistency so that people can follow that pattern when they are navigating the codebase.

I thought about it but decided not to do it that way because:

I don't think the code here is long or complex enough to need to be split up into different modules. In my opinion being in different files and modules can hurt readability (personally I don't like jumping between files), unless the code is long (transactions's case) or shareable (objects''s case), in which case the benefits outweigh the drawbacks, which bring me to the second point.

I don't see the response function in this case being shared. The only methods that return coin related responses are the methods in this file, getCoins, getAllCoins and getAllBalances. And we certainly don't plan to expand jsonrpc interface so most likely no future methods will be using it.

Or do you think this needs to be done for consistency reasons, i.e. any method that has filters and pagination should follow that structure?

Recording our offline conversation on this:

Indeed, it's mostly to do with consistency (to make it easier for people to connect the dots going from the implementation of one RpcModule to the next).

But it's true that splitting things across multiple files can hurt readability too.

Where we ended up was that we would follow the same patterns, but not split it up across files unless the file got large. In this case, with the absence of getAllCoins, this translates into:

inlining get_coins_impl back into get_coins,

renaming get_coin_id_page as filter_coins and making it a free function (in the same module),

making coin_response a free function (but in the same module).

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

crates/sui-indexer-alt-e2e-tests/tests/jsonrpc/coin/get_all_coins.snap

crates/sui-indexer-alt-jsonrpc/Cargo.toml

emmazzz · 2025-02-12T01:00:29Z

Hey @amnn thanks for the review! I have addressed most of the comments in the latest commit. Specifically,

getAllCoins has been removed. I agree with your assessment and I think we can tackle this one after Stashed milestone.
The index change has been split into its own PR [NO MERGE YET][indexer-alt] modify coin balance bucket index to sort object id in descending order #21187
I got assume_not_null working but I still need to have a separate struct for cursor for the time being because changing the object id type into Vec<u8> of the json cursor type will make it hard to test, given how we handle json cursor and object id replacement in test adapter. We are going to make it a bcs cursor anyway later when bcs cursor lands. I have added TODOs for it in the code.

I understand there are changes in other PRs that can further improve this one (bcs cursor and your improvements to load_latest) but I think these improvements are minor and don't need to block the landing of this one. What do you think?

amnn

Some small nits, but this is good to go! Agreed we can incorporate BCS cursors in a follow-up.

amnn · 2025-02-14T14:54:17Z

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

+    BadType(String, anyhow::Error),
+}
+
+// TODO: use bcs cursor once we have it.


This is now available, after a rebase, I believe.

amnn · 2025-02-14T14:54:31Z

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

+#[derive(Serialize, Deserialize, Debug, Clone)]
+struct BalanceCursor {
+    #[serde(rename = "o")]
+    object_id: ObjectID,


Suggested change

object_id: ObjectID,

object_id: Vec<u8>,

amnn · 2025-02-14T14:59:13Z

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

+    }
+}
+
+impl Coins {


Recording our offline conversation on this:

Indeed, it's mostly to do with consistency (to make it easier for people to connect the dots going from the implementation of one RpcModule to the next).

But it's true that splitting things across multiple files can hurt readability too.

Where we ended up was that we would follow the same patterns, but not split it up across files unless the file got large. In this case, with the absence of getAllCoins, this translates into:

inlining get_coins_impl back into get_coins,

renaming get_coin_id_page as filter_coins and making it a free function (in the same module),

making coin_response a free function (but in the same module).

amnn · 2025-02-14T15:01:11Z

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

+        // We get all the qualified coin ids first.
+        let coin_id_page = self.get_coin_id_page(owner, coin_type_tag, page).await?;
+
+        let coin_futures = coin_id_page.data.iter().map(|id| self.coin_response(*id));


nit: Is it possible to call into_iter here instead?

amnn · 2025-02-14T16:24:22Z

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

+                candidates.field(cb::object_id),
+                candidates.field(cb::cp_sequence_number),
+                candidates.field(cb::coin_balance_bucket).assume_not_null(),


Would the following work?

Suggested change

candidates.field(cb::object_id),

candidates.field(cb::cp_sequence_number),

candidates.field(cb::coin_balance_bucket).assume_not_null(),

candidates!(object_id),

candidates!(cp_sequence_number),

candidates!(coin_balance_bucket).assume_not_null(),

amnn · 2025-02-14T16:25:17Z

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs

+        })) = page.cursor
+        {
+            query = query.filter(
+                sql::<Bool>(r#"(candidates.coin_balance_bucket, candidates.cp_sequence_number, candidates.object_id) < ("#)


You can drop the raw string if the string contents does not include quotes:

Suggested change

sql::<Bool>(r#"(candidates.coin_balance_bucket, candidates.cp_sequence_number, candidates.object_id) < ("#)

sql::<Bool>("(candidates.coin_balance_bucket, candidates.cp_sequence_number, candidates.object_id) < (")

amnn · 2025-02-14T16:35:42Z

crates/sui-indexer-alt-e2e-tests/tests/jsonrpc/coin/pagination.move

+// This test is to verify that deleted coins are not included in the result of suix_getCoins.
+// We create two coins, of balances 12 and 34, call the rpc method to see bot of them in the results.
+// Then we merge the coins and call the rpc method again to see that only the merged coin with
+// balance 46 is in the results.
+


nit: comment copypasta

emmazzz temporarily deployed to sui-typescript-aws-kms-test-env January 21, 2025 08:39 — with GitHub Actions Inactive

vercel bot deployed to Preview – sui-docs January 21, 2025 08:40 View deployment

emmazzz force-pushed the emma/rpc-coin-etc branch from d100b09 to dbb6916 Compare January 21, 2025 18:27

emmazzz temporarily deployed to sui-typescript-aws-kms-test-env January 21, 2025 18:27 — with GitHub Actions Inactive

emmazzz marked this pull request as ready for review January 21, 2025 18:28

emmazzz requested a review from lxfind January 21, 2025 18:28

emmazzz temporarily deployed to sui-typescript-aws-kms-test-env January 21, 2025 18:28 — with GitHub Actions Inactive

emmazzz requested review from amnn, gegaowp and wlmyng January 21, 2025 18:28

vercel bot deployed to Preview – sui-docs January 21, 2025 18:28 View deployment

lxfind requested changes Jan 21, 2025

View reviewed changes

vercel bot deployed to Preview – sui-docs January 22, 2025 19:09 View deployment

emmazzz requested a review from lxfind January 22, 2025 19:12

wlmyng reviewed Jan 23, 2025

View reviewed changes

lxfind reviewed Jan 23, 2025

View reviewed changes

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs Outdated Show resolved Hide resolved

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs Outdated Show resolved Hide resolved

emmazzz force-pushed the emma/rpc-coin-etc branch from cb1f663 to e6c0d42 Compare February 3, 2025 06:58

emmazzz temporarily deployed to sui-typescript-aws-kms-test-env February 3, 2025 06:58 — with GitHub Actions Inactive

emmazzz requested review from lxfind and wlmyng February 3, 2025 06:58

vercel bot deployed to Preview – sui-docs February 3, 2025 07:00 View deployment

emmazzz force-pushed the emma/rpc-coin-etc branch from e6c0d42 to feffac2 Compare February 3, 2025 07:03

emmazzz temporarily deployed to sui-typescript-aws-kms-test-env February 3, 2025 07:03 — with GitHub Actions Inactive

vercel bot deployed to Preview – sui-docs February 3, 2025 07:05 View deployment

emmazzz commented Feb 3, 2025

View reviewed changes

crates/sui-indexer-alt-jsonrpc/src/api/coin.rs Outdated Show resolved Hide resolved

amnn reviewed Feb 3, 2025

View reviewed changes

emmazzz force-pushed the emma/rpc-coin-etc branch from feffac2 to 77055b4 Compare February 5, 2025 07:09

emmazzz temporarily deployed to sui-typescript-aws-kms-test-env February 5, 2025 07:09 — with GitHub Actions Inactive

vercel bot deployed to Preview – sui-docs February 5, 2025 08:18 View deployment

emmazzz force-pushed the emma/rpc-coin-etc branch from 2a79114 to 75d0069 Compare February 5, 2025 08:32

emmazzz temporarily deployed to sui-typescript-aws-kms-test-env February 5, 2025 08:32 — with GitHub Actions Inactive

vercel bot deployed to Preview – sui-docs February 5, 2025 08:34 View deployment

amnn reviewed Feb 5, 2025

View reviewed changes

emmazzz added 3 commits February 9, 2025 15:30

[test-adapter] refactor query interpolation and change format for bcs…

03b548d

… cursors

[indexer-alt] add getCoins jsonrpc impl

5f07a5d

address comments

7f13c3c

emmazzz force-pushed the emma/rpc-coin-etc branch from 75d0069 to 3b834a1 Compare February 10, 2025 05:55

emmazzz requested a review from stefan-mysten as a code owner February 10, 2025 05:55

emmazzz temporarily deployed to sui-typescript-aws-kms-test-env February 10, 2025 05:55 — with GitHub Actions Inactive

emmazzz changed the base branch from main to emma/txn-test-cursor February 10, 2025 05:56

vercel bot deployed to Preview – sui-docs February 10, 2025 06:00 View deployment

emmazzz commented Feb 10, 2025

View reviewed changes

emmazzz force-pushed the emma/rpc-coin-etc branch from 3b834a1 to 7d18700 Compare February 10, 2025 06:04

emmazzz temporarily deployed to sui-typescript-aws-kms-test-env February 10, 2025 06:04 — with GitHub Actions Inactive

vercel bot deployed to Preview – sui-docs February 10, 2025 06:06 View deployment

amnn reviewed Feb 11, 2025

View reviewed changes

emmazzz mentioned this pull request Feb 12, 2025

[NO MERGE YET][indexer-alt] modify coin balance bucket index to sort object id in descending order #21187

Draft

7 tasks

emmazzz added 2 commits February 11, 2025 16:47

address comments second round

1c5547b

more comments addressing

2466063

emmazzz force-pushed the emma/rpc-coin-etc branch from 7d18700 to 2466063 Compare February 12, 2025 00:50

emmazzz temporarily deployed to sui-typescript-aws-kms-test-env February 12, 2025 00:50 — with GitHub Actions Inactive

vercel bot deployed to Preview – sui-docs February 12, 2025 00:51 View deployment

emmazzz requested a review from amnn February 13, 2025 23:22

emmazzz force-pushed the emma/txn-test-cursor branch from 03b548d to b9e30f1 Compare February 14, 2025 08:43

amnn approved these changes Feb 14, 2025

View reviewed changes

Base automatically changed from emma/txn-test-cursor to main February 14, 2025 23:27

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[indexer-alt] add coin jsonrpc impl #20931

[indexer-alt] add coin jsonrpc impl #20931

emmazzz commented Jan 21, 2025 •

edited

Loading

vercel bot commented Jan 21, 2025 •

edited

Loading

lxfind Jan 21, 2025

emmazzz Jan 22, 2025

lxfind Jan 21, 2025

emmazzz Jan 22, 2025

emmazzz commented Feb 3, 2025

amnn left a comment •

edited

Loading

amnn Feb 3, 2025

emmazzz Feb 3, 2025

amnn left a comment

amnn Feb 5, 2025

amnn Feb 5, 2025

amnn Feb 5, 2025

emmazzz Feb 10, 2025

amnn Feb 11, 2025

emmazzz commented Feb 10, 2025

amnn left a comment

amnn Feb 11, 2025

emmazzz Feb 11, 2025 •

edited

Loading

amnn Feb 14, 2025

emmazzz commented Feb 12, 2025

amnn left a comment

amnn Feb 14, 2025

amnn Feb 14, 2025

amnn Feb 14, 2025

amnn Feb 14, 2025

amnn Feb 14, 2025

amnn Feb 14, 2025

amnn Feb 14, 2025

	if let Some(c) = page.cursor {
	query = query.filter(
	sql::<Bool>(r#"("candidates"."cp_sequence_number", "candidates"."object_id") < ("#)
	.bind::<BigInt, _>(c.cp_sequence_number as i64)
	.sql(", ")
	.bind::<Bytea, _>(c.object_id.clone())
	.sql(")"),
	);
	}

	sql::<Bool>(r#"(candidates.coin_balance_bucket, candidates.cp_sequence_number, candidates.object_id) < ("#)
	sql::<Bool>("(candidates.coin_balance_bucket, candidates.cp_sequence_number, candidates.object_id) < (")

[indexer-alt] add coin jsonrpc impl #20931

Are you sure you want to change the base?

[indexer-alt] add coin jsonrpc impl #20931

Conversation

emmazzz commented Jan 21, 2025 • edited Loading

Description

Test plan

Release notes

vercel bot commented Jan 21, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

emmazzz commented Feb 3, 2025

amnn left a comment • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

amnn left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

emmazzz commented Feb 10, 2025

amnn left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

emmazzz Feb 11, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

emmazzz commented Feb 12, 2025

amnn left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

emmazzz commented Jan 21, 2025 •

edited

Loading

vercel bot commented Jan 21, 2025 •

edited

Loading

amnn left a comment •

edited

Loading

emmazzz Feb 11, 2025 •

edited

Loading