The match protocol can be quite long and it may be desirable to search for a specific entry, e.g. by protocol type, game event, command, etc.

Such filters could be added to the protocol view (not the side bar).

This issue lists Renovate updates and detected dependencies. Read the Dependency Dashboard docs to learn more.

Awaiting Schedule

These updates are awaiting their schedule. Click on a checkbox to get an update now.

• Update cimg/base Docker tag to v2024.05
• Update cimg/go Docker tag to v1.22.3
• Update golang Docker tag to v1.22
• Update module github.com/google/uuid to v1.6.0
• Update module google.golang.org/protobuf to v1.34.1
• Update robocupssl/ssl-log-player Docker tag to v1.6.0
• Update robocupssl/ssl-remote-control Docker tag to v1.3.0
• Update robocupssl/ssl-status-board Docker tag to v2.11.1
• Update tigersmannheim/auto-referee Docker tag to v1.3.0

Detected dependencies

• Check this box to trigger a request for Renovate to run again on this repository

Currently there's no documentation for a large part of the fields of what units they have for GameEvent, particularly the locations. Are these in meters? or millimiters (ssl-vision standard?)?

Find a Sumatra Recording attached:

2024-01-30_13-33-02.zip

What happens in the recording? Note both AIs are set to off.

1. The ball is placed manually inside the blue defense area.
2. After 5s a Keeper Held Ball Foul is issued
3. The ball placement is done manually by just placing the ball
4. After the successful ball placement the game is continued with a force start
5. The ball is placed back into the defense area
6. After 5s a Keeper Held Ball Foul is issued
7. Yet again ball placement, but this time the ball placement is interrupted with a manual force start
8. The ball is placed back into the defense area
9. Now it takes more than 20s for the Keeper Held Ball Foul to be issued.

Note the 30s ball placement timer is still counting down, although the ball placement was already aborted with the force start.

If I start the game controller as described in the readme.md, but the computer doesn't have an active internet connection (i.e. wifi disabled, no LAN), the game controller crashes with the following stack trace:

panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x68 pc=0x78f4f0]

goroutine 26 [running]:
github.com/RoboCup-SSL/ssl-game-controller/internal/app/controller.(*RefMessage).setState(0xc420080908, 0xc42018c200, 0x431ce0)
/home/andi/go/src/github.com/RoboCup-SSL/ssl-game-controller/internal/app/controller/publisher.go:106 +0x70
github.com/RoboCup-SSL/ssl-game-controller/internal/app/controller.(*RefMessage).Publish(0xc420080908, 0xc42018c200)
/home/andi/go/src/github.com/RoboCup-SSL/ssl-game-controller/internal/app/controller/publisher.go:85 +0x35
github.com/RoboCup-SSL/ssl-game-controller/internal/app/controller.(*Publisher).Publish(0xc420080900, 0xc42018c200)
/home/andi/go/src/github.com/RoboCup-SSL/ssl-game-controller/internal/app/controller/publisher.go:80 +0x3b
github.com/RoboCup-SSL/ssl-game-controller/internal/app/controller.(*GameController).publishNetwork(0xc420080800)
/home/andi/go/src/github.com/RoboCup-SSL/ssl-game-controller/internal/app/controller/controller.go:251 +0x5e
created by github.com/RoboCup-SSL/ssl-game-controller/internal/app/controller.(*GameController).Run
/home/andi/go/src/github.com/RoboCup-SSL/ssl-game-controller/internal/app/controller/controller.go:215 +0x1a3
exit status 2

OS: Ubuntu20.04

The error is following

$ ./ssl-game-controller_v3.0.0_linux_amd64 
./ssl-game-controller_v3.0.0_linux_amd64: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.32' not found (required by ./ssl-game-controller_v3.0.0_linux_amd64)
./ssl-game-controller_v3.0.0_linux_amd64: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.34' not found (required by ./ssl-game-controller_v3.0.0_linux_amd64)

Since freekicks were merged into just direct free kicks in 2021, it would make sense to merge the buttons in the gamecontroller as well for clarity.

Context: Recently, we tried importing the new protobuf definitions from this repo into our repo, namely ssl_gc_common.proto, ssl_gc_game_event.proto, ssl_gc_geometry.proto, and ssl_gc_referee_message.proto. We wanted to do this because https://github.com/RoboCup-SSL/ssl-refbox/ has been deprecated, and this repo's definitions has some updated documentation.

Problem: We encountered name conflicts between our internal Team and RobotId classes and the ones defined in ssl_gc_common.proto. This is a change from defining them inside the GameEvent message as in https://github.com/RoboCup-SSL/ssl-refbox/. We also noticed that SSL_Referee has been renamed to Referee, which might cause name conflicts in the future (but currently don't in our repo). We're curious why these changes were made as it causes problems for easily updating these protobufs in our code and other teams would probably encounter the same issues.

Suggested solution: Prefix protobuf messages with SSL_ or embed them in other message definitions, as was done in https://github.com/RoboCup-SSL/ssl-refbox/

Problem

Ball placement can fail for different reasons. The most obvious one is, when the ball is not yet on the placement position. If it is, the placing robot still needs to bring some distance to the ball, see: https://robocup-ssl.github.io/ssl-rules/sslrules.html#_ball_placement
The PLACEMENT_FAILED event only includes the remaining_distance. If it is close to zero, it is not obvious, for which reason the placement failed.

Current situation

Currently, the autoRefs are responsible to create PLACEMENT_SUCCESSFUL events, when all conditions are met.
The GC will create a PLACEMENT_FAILED event, when no successful event was sent by the autoRefs in time.
The GC will include the remaining distance from ball to placement position in the failed event.

Proposed solution

Easy short-term solution

• Add a nearest_own_bot_distance property to the PLACEMENT_FAILED event.
• Calculate the value in the GC and add it to the event

Better solution

The above proposed solution does not solve the problem reliably: There could still be an inconsistency between how an autoRef judges the success / failure of the placement. Specifically, if using majority voting and a majority is not reached, the placement would be counted as failed, without reporting additional information about why it failed.

A better solution could be to report success and failure consistently at only one place: Either the autoRefs or the GC:

• The responsibility of sending the PLACEMENT_FAILED event could be transferred to the autoRefs. However, especially with the current majority implementation, there is no guaranty, that all autoRefs report the failure, so the timeout mechanism of placement may not work.
• The responsibility of sending the PLACEMENT_SUCCESSFUL event could be transferred to the GC. The majority mechanism would be lost. However, the detection is not that complicated. This could also solve another corner case: If multiple autoRefs report the success at different times, the current majority implementation would not accept it.

I was testing with the latest release (2.0.7)

To reproduce:
-Run the gamecontroller without vision available:
-Run a penalty (Stop->Penalty->NormalStart)
-Let the time (10 seconds) run out

After this happens, we get the following error on our system receiving the referee messages:

[libprotobuf ERROR google/protobuf/message_lite.cc:122] Can't parse message of type "proto.Referee" because it is missing required fields: game_events[0].type

In any case, I don't think the GameController should send unparseable messages so I think this is a valid issue.

The num-Pad Shortcuts can not be disabled temporary, so if a number is entered on the num-pad somewhere, this will also trigger the assigned commands.

Either need to change the shortcuts, or have a look into the shortcut library.