Zenoh 1.0.0 Porting #276

YuanYuYuan · 2024-09-05T13:17:03Z

Notes

Remove some unused dependencies.
Use z_view_keyexpr_t rather z_owned_keyexpr_t if possible. This keeps the reference only and doesn't need to be dropped.
The attachment type has been removed. Use the plain z_bytes instead.
Attachment is nothing but z_bytes. Use an iterator to get value by key.
z_check is now internal. We check zenoh entities' health by the creation functions' return.

Signed-off-by: Luca Cominardi <[email protected]>

chore: checkout dev/1.0.0

…n the zenoh_c_vendor

Yadunund

Re-reviewed 75% of the changes and provided some feedback. My primary concern is the introduction of registering a function with the atexit signal handler. As explained inline, there is still a problem of the static variable being invalidated and i'm confused why we need it with dev/1.0.0 when we do not need it rolling.

Secondly, I think the diff in the PR is way larger than it needs to be. Imo the changes only need reflect changes in the zenoh-c API (and things like attachment_helpers) but there are changes in rmw_zenoh.cpp wrt to checking if entities are valid, zenoh utils, and various cleanups to headers.
My recommendation is to take all the changes that remove cleanup headers and target a PR to the current rolling branch. And save other changes unrelated to breaking zenoh-c API for a future PR.

rmw_zenoh_cpp/src/zenohd/main.cpp

Yadunund · 2024-11-23T18:24:49Z

rmw_zenoh_cpp/src/detail/rmw_client_data.cpp

@@ -361,28 +351,21 @@ rmw_ret_t ClientData::take_response(
  }

  // Fill in the request_header
-  request_header->request_id.sequence_number =
-    rmw_zenoh_cpp::get_int64_from_attachment(&sample->attachment, "sequence_number");
+  rmw_zenoh_cpp::attachment_data_t attachment(z_sample_attachment(sample));


This API feels awkward and requires you to define the move constructor. Why not just pass const z_loaned_sample_t * and construct the z_sample_attachment internally?

Sorry to miss this. The attachment is nothing but the usual zenoh bytes in 1.0.
On the other hand, zenoh-c basically doesn't provide any special structure for the users.
That's why we need to define an attachment data type and its (de)serialization.

The rmw_zenoh_cpp::attachment_data_t in this PR is considered to mapped to a z_bytes. One of the constructor is defined as

attachment_data_t::attachment_data_t(const z_loaned_bytes_t * attachment) { ... }

which means we can parse the given attachment from raw zenoh bytes to the desired data types.

We also use this pattern in zenoh-cpp.

auto attachment = sample.get_attachment(); if (!attachment.has_value()) return; // Assume we convert the raw bytes to a `std::unordered_map<std::string, std::string>` auto attachment_data = ext::deserialize<std::unordered_map<std::string, std::string>>(attachment->get());

Yadunund · 2024-11-23T18:28:47Z

rmw_zenoh_cpp/src/detail/rmw_client_data.cpp

-  if (zc_liveliness_token_check(&token_)) {
-    zc_liveliness_undeclare_token(z_move(token_));
-  }
-  if (z_check(z_loan(keyexpr_))) {
-    z_drop(z_move(keyexpr_));
-  }
+  zc_liveliness_undeclare_token(z_move(token_));
+
+  z_drop(z_move(keyexpr_));


We previously needed this check since it was possible for token_ and keyexpr_ to be uninitialized when shutdown is called within ClientData::make. Do these functions now check whether the tokens are valid before dropping?

Actually, this is one of the misunderstandings we would like to avoid.

There's no way to check an uninitialized zenoh entity.

z_owned_keyexpr_t keyexpr; z_check(z_loan(keyexpr)) // This is an undefined behavior.

A proper way to use z_check is to use it on an initialized data.

z_owned_keyexpr_t keyexpr; z_keyexpr_from_str(&keyexpr, "foo"); z_check(z_loan(keyexpr)); // This is okay. z_owned_keyexpr_t keyexpr; z_null(&keyexpr); z_check(z_loan(keyexpr)); // This is also fine.

The check itself checks the gravestone status of zenoh data. It doesn't do functional checks like liveliness token querying over the network. These functions are necessary for zenoh-cpp development yet are easy to misuse. That's why we finally decided to make them internal. See z_check and z_null made internal eclipse-zenoh/zenoh-c#605 for details.

Regarding the contention issue you mentioned, I assume that the function ClientData::shutdown can only be called once a valid ClientData instance has been created. However, if we prefer to relax this restriction, we could introduce a flag to indicate whether the class object has been properly validated. This would help address the issue. What are your thoughts?

Thanks for explaining that. Right now ClientData::shutdown can be invoked even if the tokens are not properly initialized. So it sounds like we would need to introduce an initialized_ flag to the implementation.

I've uniformly applied the initialized_ flag on all the detail/rmw_*_data in ZettaScaleLabs@4c63e8d

rmw_zenoh_cpp/src/detail/rmw_context_impl_s.cpp

rmw_zenoh_cpp/src/rmw_zenoh.cpp

Yadunund · 2024-11-23T19:26:23Z

rmw_zenoh_cpp/src/rmw_zenoh.cpp

-  if (!context_impl->session_is_valid()) {
-    RMW_SET_ERROR_MSG("zenoh session is invalid");
-    return nullptr;
-  }


Why remove this?

Short answer: the z_session_check function has been removed. The z_check function has been changed to internal in eclipse-zenoh/zenoh-c#605. The check itself is to check the gravestone health of the zenoh data. If we want to check whether the session is constructed successfully, we should use the return code during the creation.

Let's say the session was constructed successfully but was closed later on. How would we check if the session is still open?

There's a function z_session_is_closed and it's been used as the implementation of rmw_context_impl_s::Data session_is_valid. I think I can add it back to this.

Addressed in ZettaScaleLabs@09b45c3.

rmw_zenoh_cpp/src/rmw_zenoh.cpp

Yadunund · 2024-11-23T19:27:14Z

rmw_zenoh_cpp/src/rmw_zenoh.cpp

-  if (!pub_data->liveliness_is_valid()) {
-    return RMW_RET_ERROR;
-  }
-


Why remove this?

As stated above, the z_check has been removed.

~~Addressed in ZettaScaleLabs@09b45c3.~~

Addressed in ZettaScaleLabs@84d1267.

Yadunund · 2024-11-23T19:27:40Z

rmw_zenoh_cpp/src/rmw_zenoh.cpp

-  if (!context_impl->session_is_valid()) {
-    RMW_SET_ERROR_MSG("zenoh session is invalid");
-    return nullptr;
-  }


Again, why remove this here and in other functions below?

As stated above, the z_check has been removed.

Addressed in ZettaScaleLabs@09b45c3.

* Fix `z_view_string_t` to `std::string` conversion * Fix formatting

YuanYuYuan · 2024-12-02T18:15:17Z

Latency comparison

ros2-simple-performance: ZettaScaleLabs/ros2-simple-performance@afe1820
warm up time (sec): 4
samples number: 5000
publish rate (hz): 500
qos: reliable,volatile,keep_all
latency: rtt (us)
branch
- rolling (based on zenoh 0.11): 8306a63
- session (based on zenoh 1.0): ZettaScaleLabs@f5efd02

branch	payload	min	p05	p50	p95	max
rolling	32b	229	247	430	514	2850
session	32b	187	195	219	438	4219
rolling	64k	422	443	625	742	4535
session	64k	203	215	232	459	4729

Conclusion: the rmw_zenoh 1.0 is almost twice as fast as the 0.11.0 impl.
it can be further improved by make z_bytes_to_slice and z_bytes_to_string avoid making copy for non-fragmented non-shm data eclipse-zenoh/zenoh-c#822 . Stay tuned!

YuanYuYuan · 2024-12-02T19:28:01Z

Pacakge	Test	rolling	session	dev/1.0	Note
rclcpp	18 - test_allocator_memory_strategy (Timeout)			X
rclcpp	23 - test_clock			X
rclcpp	28 - test_client_common			X	#328
rclcpp	62 - test_publisher	X	X	X	Identified
rclcpp	91 - test_wait_for_message			X
rclcpp	95 - test_executors		X	X
rclcpp	99 - test_executors_busy_waiting	F		X
rclcpp	118 - test_executor (Timeout)		F
rclcpp_action	1 - test_client	X	X	X	#326
rclcpp_action	2 - test_server		X	X	severe
rclcpp_lifecycle	6 - test_lifecycle_node		F	?
rclcpp_lifecycle	9 - test_lifecycle_service_client	X	X	?	To be checked
test_communication	73 - test_publisher_subscriber__rclpy__rmw_zenoh_cpp			X
test_communication	78 - test_action_client_server__rclpy__rclcpp__rmw_zenoh_cpp			X
test_communication	81 - test_action_client_server__rclcpp__rclpy__rmw_zenoh_cpp			X
test_communication	83 - test_requester_replier__rclcpp__rmw_zenoh_cpp		X	X	Related to #329
test_communication	84 - test_action_client_server__rclcpp__rmw_zenoh_cpp			X
test_rclcpp	70 - test_parameter_server_cpp__rmw_zenoh_cpp (Timeout)		X	X	To be checked
test_rclcpp	74 - test_n_nodes__rmw_zenoh_cpp		F	F	#329

F: Flaky
rolling (based on zenoh 0.11): 8306a63
dev/1.0 (based on zenoh 1.0): ZettaScaleLabs@435186a
session (based on zenoh 1.0): ZettaScaleLabs@f5efd02, wrapping the zenoh session with std::shared_ptr

YuanYuYuan · 2024-12-04T17:56:47Z

Follow-up on Review Feedback

Remove the header clean.
Remove unneeded files.
Separate the atexit handler hack. The solution proposal for panic in the program termination #324.
Reorder the scope_exit. Previously, we have

z_owned_publisher_t pub;
auto free_type_hash_c_str = rcpputils::make_scope_exit([&]() {
    z_drop(z_move(pub));
});
if (z_declare_publisher(session, &pub, ...) != Z_OK) {
    return;
}

Since we actually don't need to z_drop the zenoh entity if we fail to establish it. This can be reordered in a safer way in case we drop an uninitialized memory.

z_owned_publisher_t pub;
if (z_declare_publisher(session, &pub, ...) != Z_OK) {
    return;
}
auto free_type_hash_c_str = rcpputils::make_scope_exit([&]() {
    z_drop(z_move(pub));
});

(*) Sharing the zenoh session with a referece counter is proposed. Issues caused by not shutting down the contained entity ZettaScaleLabs/rmw_zenoh#44.
(*) Adding initialized_ flag to guard the abrupt shutdown of all rmw_*_data.
(Optional) The result of z_liveliness_undeclare_token(z_move(token_)) in ClientData::_shutdown is not addressed.

(*) This means the change is held on the branch https://github.com/ZettaScaleLabs/rmw_zenoh/tree/wip/session-clone and not yet merged into dev/1.0.0.

clalancette · 2024-12-04T19:54:03Z

All right, so I just tried out the latest here (I'm sorry I haven't gotten to this earlier).

First of all, I needed the following patch to compile:

diff --git a/rmw_zenoh_cpp/src/detail/rmw_context_impl_s.cpp b/rmw_zenoh_cpp/src/detail/rmw_context_impl_s.cpp
index 3c89983..ed3bd9a 100644
--- a/rmw_zenoh_cpp/src/detail/rmw_context_impl_s.cpp
+++ b/rmw_zenoh_cpp/src/detail/rmw_context_impl_s.cpp
@@ -152,10 +152,10 @@ public:
         const z_loaned_sample_t * sample = z_reply_ok(z_loan(reply));
         z_view_string_t keystr;
         z_keyexpr_as_view_string(z_sample_keyexpr(sample), &keystr);
-        std::string livelines_str(z_string_data(z_loan(keystr)), z_string_len(z_loan(keystr)));
+        std::string liveliness_str(z_string_data(z_loan(keystr)), z_string_len(z_loan(keystr)));
         // Ignore tokens from the same session to avoid race conditions from this
         // query and the liveliness subscription.
-        graph_cache_->parse_put(std::move(livelines_str), true);
+        graph_cache_->parse_put(std::move(liveliness_str), true);
       } else {
         RMW_ZENOH_LOG_DEBUG_NAMED(
           "rmw_zenoh_cpp", "[rmw_context_impl_s] z_call received an invalid reply.\n");
@@ -439,8 +439,8 @@ static void graph_sub_data_handler(z_loaned_sample_t * sample, void * data)
   }
 
   // Update the graph cache.
-  std::string livelines_str(z_string_data(z_loan(keystr)), z_string_len(z_loan(keystr)));
-  data_shared_ptr->update_graph_cache(z_sample_kind(sample), std::move(livelines_str));
+  std::string liveliness_str(z_string_data(z_loan(keystr)), z_string_len(z_loan(keystr)));
+  data_shared_ptr->update_graph_cache(z_sample_kind(sample), std::move(liveliness_str));
 }
 
 ///=============================================================================
diff --git a/rmw_zenoh_cpp/src/detail/rmw_publisher_data.cpp b/rmw_zenoh_cpp/src/detail/rmw_publisher_data.cpp
index cbc7e08..5966de6 100644
--- a/rmw_zenoh_cpp/src/detail/rmw_publisher_data.cpp
+++ b/rmw_zenoh_cpp/src/detail/rmw_publisher_data.cpp
@@ -173,8 +173,7 @@ std::shared_ptr<PublisherData> PublisherData::make(
   z_view_keyexpr_from_str(&liveliness_ke, liveliness_keyexpr.c_str());
   z_owned_liveliness_token_t token;
   if (z_liveliness_declare_token(
-      z_loan(owned_session), &token, z_loan(liveliness_ke),
-      NULL) != Z_OK)
+      session, &token, z_loan(liveliness_ke), NULL) != Z_OK)
   {
     RMW_ZENOH_LOG_ERROR_NAMED(
       "rmw_zenoh_cpp",

After that, I tried to run the tests in rcl, and I'm getting a lot of crashing tests:

The following tests FAILED:
	  9 - test_context__rmw_zenoh_cpp (Failed)
	 15 - test_init__rmw_zenoh_cpp (Failed)
	 16 - test_node__rmw_zenoh_cpp (Failed)
	 19 - test_guard_condition__rmw_zenoh_cpp (Failed)
	 37 - test_time (Failed)

I also tried out the wip/session branch, and with that one I'm seeing about the same crashing tests. So something is going on there that I need to investigate. I'm spending some time on it now, but I thought I'd put a brief report on what I'm seeing first.

clalancette · 2024-12-04T21:09:38Z

After that, I tried to run the tests in rcl, and I'm getting a lot of crashing tests:

Ah, never mind, this was a local problem. Things are working OK now (though I still need the patch above). Will continue testing.

Yadunund · 2024-12-05T00:54:16Z

rmw_zenoh_cpp/src/detail/rmw_client_data.cpp

-{
-  std::lock_guard<std::recursive_mutex> lock(mutex_);
-  return zc_liveliness_token_check(&token_);
-}


I understand that these z_check functions have been made internal but having a method to check whether the liveliness token is still valid is useful. How about we simply return !is_shutdown_ here since we unregister the token only after the entity is shutdown. Then keep the function call in rmw_zenoh.cpp.

It seems that we only check liveliness_is_valid in rmw_zenoh_cpp/src/rmw_zenoh.cpp. Let me change it with is_shutdown() there. ZettaScaleLabs@84d1267

rolling branch

rmw_zenoh_cpp/src/rmw_zenoh.cpp 754: if (!pub_data->liveliness_is_valid()) { rmw_zenoh_cpp/src/detail/rmw_service_data.hpp 63: bool liveliness_is_valid() const; rmw_zenoh_cpp/src/detail/rmw_service_data.cpp 242:bool ServiceData::liveliness_is_valid() const rmw_zenoh_cpp/src/detail/rmw_publisher_data.cpp 431:bool PublisherData::liveliness_is_valid() const rmw_zenoh_cpp/src/detail/rmw_publisher_data.hpp 74: bool liveliness_is_valid() const; rmw_zenoh_cpp/src/detail/rmw_subscription_data.hpp 90: bool liveliness_is_valid() const; rmw_zenoh_cpp/src/detail/rmw_client_data.cpp 281:bool ClientData::liveliness_is_valid() const rmw_zenoh_cpp/src/detail/rmw_subscription_data.cpp 389:bool SubscriptionData::liveliness_is_valid() const rmw_zenoh_cpp/src/detail/rmw_client_data.hpp 64: bool liveliness_is_valid() const;

…c version

YuanYuYuan · 2024-12-05T07:09:47Z

After that, I tried to run the tests in rcl, and I'm getting a lot of crashing tests:

Ah, never mind, this was a local problem. Things are working OK now (though I still need the patch above). Will continue testing.

The branch is somewhat out of sync with the wip/session-clone. I just pushed a fix with the typo fix. Thanks!

YuanYuYuan · 2024-12-05T15:50:47Z

New results

Pacakge	Test	Session	Rolling
rclc	1 - rclc_test (Failed)	X
rclc_parameter	1 - rclc_parameter_test (Timeout)	X	X
rclcpp	118 - test_executor (Timeout)	X
rclcpp	28 - test_client_common (Failed)	X
rclcpp	62 - test_publisher (Failed)	X	X
rclcpp	86 - test_time (Failed)		X
rclcpp	96 - test_executors_timer_cancel_behavior (Failed)		X
rclcpp_action	1 - test_client (Failed)	X	X
rclcpp_action	2 - test_server (Failed)	X
rclcpp_lifecycle	9 - test_lifecycle_service_client (Failed)	X	X
rclpy	7 - test_client (Failed)		X
test_communication	83 - test_requester_replier__rclcpp__rmw_zenoh_cpp (Failed)	X
test_rclcpp	70 - test_parameter_server_cpp__rmw_zenoh_cpp (Timeout)	X
test_rclcpp	74 - test_n_nodes__rmw_zenoh_cpp (Failed)	X

Session: ZettaScaleLabs@1ecdd5c
Rolling: 8306a63
Some test_quality_of_service failures are omitted.

Signed-off-by: Yadunund <[email protected]>

Stale review.

Mallets and others added 30 commits August 9, 2024 14:32

Update style.yaml

e2fabb6

Signed-off-by: Luca Cominardi <[email protected]>

Update check_logging.yaml

6cd6dbe

Signed-off-by: Luca Cominardi <[email protected]>

Update build.yaml

dcefb68

Signed-off-by: Luca Cominardi <[email protected]>

chore: configure the compiliation

14fe58c

chore: checkout dev/1.0.0

chore: complete the 1st version

301a8ce

fix: memory leak

8ef92b5

wip

06b6fa4

fix: z_error_t -> z_result_t

71e7c5b

Fix scouting/*/autoconnect/* per eclipse-zenoh/zenoh@b31a410 (#3)

a717270

chore: checkout the local zenoh-c

cfa63d9

chore: polish z_open

12dc820

fix: segfault

d331b74

feat: z_bytes_serialize_from_slice without copy

1db4bef

chore: switch back to https://github.com/eclipse-zenoh/zenoh-c

d12f189

Initialize query_ member of ZenohQuery

909329b

refactor: use z_owned_slice_t instead

511c6f6

chore: adapt the latest change of zenoh-c dev/1.0.0

044d25b

chore: use strncmp to avoid copying

d79cf5e

refactor: use z_view_keyexpr_t to avoid copying

6637400

feat: implment the SHM feature

83334c8

chore: adpat the new changes from zenoh-c and fix the bug in liveliness

bf0da8a

fix: segmentation fault due to the unallocated query memory

dfc95ad

fix: workaround the ZID parsing issue

cd071e5

fix Zenoh Config read\check

4f842fd

adopt to recent zenoh-c API changes

af2e344

chore: sync up with the rolling branch

0225a59

chore: update to self-hosted runners

36140fd

fix: adapt the latest change of batching config

080eb15

Merge remote-tracking branch 'origin/rolling' into dev/1.0.0

f60c2a2

build: deprecate the zenohc_debug and incldue the zenohc dependency i…

83a26f2

…n the zenoh_c_vendor

Yadunund requested changes Nov 23, 2024

View reviewed changes

fuzzypixelz and others added 6 commits November 27, 2024 19:38

Fix z_view_string_t to std::string conversion (#45)

0dffa49

* Fix `z_view_string_t` to `std::string` conversion * Fix formatting

fix: address the comments

91df63c

style: ament_cpplint

70990e7

refactor: remove the workaround for the atexit shutdown

33f0bec

fix: correct the call by ref

b8ce65c

refactor: revert the header clean

435186a

YuanYuYuan mentioned this pull request Dec 3, 2024

Issues caused by not shutting down the contained entity ZettaScaleLabs/rmw_zenoh#44

Closed

YuanYuYuan added 4 commits December 5, 2024 01:44

refactor: zc_liveliness_* -> z_liveliness_* and bump up zenoh-c version

93afc16

refactor: reorder the cancel functions

17865d2

chore: reorder some lines of code

0255c37

refactor: add session_is_valid check

09b45c3

fixup! refactor: reorder the cancel functions

0e1cca1

Yadunund reviewed Dec 5, 2024

View reviewed changes

YuanYuYuan added 2 commits December 5, 2024 15:05

fixup! refactor: zc_liveliness_* -> z_liveliness_* and bump up zenoh-…

09ecde7

…c version

fixup! fix: address the comments

8f4949c

chore: address the comment

84d1267

Minor tweaks

08b6f53

Signed-off-by: Yadunund <[email protected]>

Yadunund approved these changes Dec 6, 2024

View reviewed changes

clalancette merged commit 0293278 into ros2:rolling Dec 6, 2024
6 checks passed

YuanYuYuan mentioned this pull request Dec 11, 2024

Porting rmw_zenoh_cpp to zenoh 1.0 ZettaScaleLabs/rmw_zenoh#15

Closed

4 tasks

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Zenoh 1.0.0 Porting #276

Zenoh 1.0.0 Porting #276

YuanYuYuan commented Sep 5, 2024 •

edited

Loading

Yadunund left a comment •

edited

Loading

Yadunund Nov 23, 2024

YuanYuYuan Dec 4, 2024

Yadunund Nov 23, 2024

YuanYuYuan Nov 27, 2024

Yadunund Dec 3, 2024

YuanYuYuan Dec 4, 2024 •

edited

Loading

Yadunund Nov 23, 2024

YuanYuYuan Nov 27, 2024 •

edited

Loading

Yadunund Nov 27, 2024

YuanYuYuan Dec 3, 2024

YuanYuYuan Dec 4, 2024 •

edited

Loading

Yadunund Nov 23, 2024

YuanYuYuan Nov 28, 2024

YuanYuYuan Dec 4, 2024 •

edited

Loading

Yadunund Nov 23, 2024

YuanYuYuan Nov 28, 2024

YuanYuYuan Dec 4, 2024 •

edited

Loading

YuanYuYuan commented Dec 2, 2024 •

edited

Loading

YuanYuYuan commented Dec 2, 2024 •

edited

Loading

YuanYuYuan commented Dec 4, 2024 •

edited

Loading

clalancette commented Dec 4, 2024

clalancette commented Dec 4, 2024

Yadunund Dec 5, 2024

YuanYuYuan Dec 5, 2024 •

edited

Loading

YuanYuYuan commented Dec 5, 2024

YuanYuYuan commented Dec 5, 2024

Zenoh 1.0.0 Porting #276

Zenoh 1.0.0 Porting #276

Conversation

YuanYuYuan commented Sep 5, 2024 • edited Loading

Notes

Yadunund 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

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

YuanYuYuan Dec 4, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

YuanYuYuan Nov 27, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

YuanYuYuan Dec 4, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

YuanYuYuan Dec 4, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

YuanYuYuan Dec 4, 2024 • edited Loading

Choose a reason for hiding this comment

YuanYuYuan commented Dec 2, 2024 • edited Loading

Latency comparison

YuanYuYuan commented Dec 2, 2024 • edited Loading

YuanYuYuan commented Dec 4, 2024 • edited Loading

Follow-up on Review Feedback

clalancette commented Dec 4, 2024

clalancette commented Dec 4, 2024

Choose a reason for hiding this comment

YuanYuYuan Dec 5, 2024 • edited Loading

Choose a reason for hiding this comment

YuanYuYuan commented Dec 5, 2024

YuanYuYuan commented Dec 5, 2024

New results

YuanYuYuan commented Sep 5, 2024 •

edited

Loading

Yadunund left a comment •

edited

Loading

YuanYuYuan Dec 4, 2024 •

edited

Loading

YuanYuYuan Nov 27, 2024 •

edited

Loading

YuanYuYuan Dec 4, 2024 •

edited

Loading

YuanYuYuan Dec 4, 2024 •

edited

Loading

YuanYuYuan Dec 4, 2024 •

edited

Loading

YuanYuYuan commented Dec 2, 2024 •

edited

Loading

YuanYuYuan commented Dec 2, 2024 •

edited

Loading

YuanYuYuan commented Dec 4, 2024 •

edited

Loading

YuanYuYuan Dec 5, 2024 •

edited

Loading