When a client component (e.g. the menu or the console) disconnects the client, this immediately unloads the map data but the render call continues as normal. This causes all components rendered after this component to access invalid memory in place of the map, layers and collision data.
This is fixed by delaying the actual disconnecting until after the render call, to ensure that the map data can be safely unloaded.
Closes#6387. Closes#3179.
To show an error/warning/information message popup with variable title and message.
This uses the SDL function `SDL_ShowSimpleMessageBox` to show the message box, because it is simpler than implementing this ourself in the base system, especially because we would have to add an additional explicit dependency on GTK3 to show a message dialog on Linux.
This function can be used without SDL being initialized.
6541: Add the DDNet git hash and "is steam" to the crash file r=Robyt3 a=Jupeyy
`@def-` what do you think, how hard would it be for the symbols uploaded to the website contain the same git hash in the name, so we could write a script that automatically finds the correct binary and creates a stack trace?
## Checklist
- [ ] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test (especially base/) or added coverage to integration test
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: Jupeyy <jupjopjap@gmail.com>
The `Disconnect` function did not fully clear all server data, because the client state was already set to offline manually before the function is called. This was causing inconsistent behavior when connecting to another server after being disconnected. For example, the client could get stuck at "getting game info" when connecting the next time.
Port the line input (UI edit boxes, chat, console) and Input Method Editor (IME) support from upstream. Closes#4397.
General
------------------------------
Fix issues with the text input. Closes#4346. Closes#4524.
Word skipping (when holding Ctrl) is overhauled to be consistent with the Windows / Firefox experience that I took as reference.
Improve usability by not blinking (i.e. always rendering) the caret shortly after is has been moved.
UI text input
------------------------------
Fix inconsistent mouse-based left and right scrolling (closes#4347).
Support smooth left and right scrolling.
Chat
------------------------------
Support keyboard-based text selection of the chat input.
Mouse-based selection could be support in the future when we decide to add something like an ingame UI cursor.
Support smooth up and down scrolling of the chat input, removing the old hack that offsets the input string to simulate scrolling.
Console
------------------------------
Also support mouse-based text selection of the command input.
Only text from either the command input or the console log can be selected at the same time. This ensures that Ctrl+C will always copy the text that is currently visually selected in the console.
Check for Ctrl+C input event in event handler instead of in render function, to hopefully fix the issue that copying does not work sometimes (closes#5974 until further notice).
When Ctrl+C is used to copy text from the console log, the selection is cleared. This should make it more clear when text was copied from the log.
Fix an issue that was preventing the console log selection from being cleared, when all log lines are selected.
Remove Ctrl+A/E hotkeys that move cursor to beginning/end respectively. Ctrl+A now selectes all text like for all other inputs. Home and End keys can still be used to go the beginning and end.
Remove Ctrl+U/K hotkeys that clear everything before/after the cursor respectively. Hold shift and use Home/End to select everything instead.
IME support
------------------------------
Render list of IME candidates in the client on Windows, so the candidate list can also be viewed in fullscreen mode. There is no API available to retrieve a candidate list on the other operating systems.
Improve composition rendering by underlining the composition text instead of putting it in square brackets.
Track active input globally to properly activate and deactivate IME through the SDL functions.
Closes#1030. Closes#1008.
Password rendering
------------------------------
Fix rendering of passwords containing unicode. Instead of rendering one star character for each UTF-8 `char`, render on star for every unicode codepoint.
Show the composition text also for passwords. Without seeing the composition text it's hard to type a password containing those characters. The candidate window exposes the composition anyway. If you don't want to expose your password this way, e.g. while streaming, you could:
1. Use a latin password and switch off the IME for the password input with the IME hotkey.
2. Blank your screen with an external program while you are streaming and entering passwords.
3. Use binds to authenticate in rcon or to set the server browser password.
Refactoring
------------------------------
Move all text input logic and general rendering to `CLineInput`.
A `CLineInput` is associated with a particular `char` buffer given as a pointer either in the constructor or with `SetBuffer`. The maximum byte size of the buffer must also be specified. The maximum length in unicode codepoints can also be specified separately (e.g. on upstream, name are limited by the number of unicode codepoints instead).
Add `CLineInputBuffered`, which is a `CLineInput` that own a `char` buffer of a fixed size, which is specified as a template argument. As `CLineInput` does not own a buffer anymore, this reduces duplicate code for line inputs that need their own buffer.
Add `CLineInputNumber` which has additional convenience functions to consider the text as an `int` or `float`, to reduce duplicate code in those cases. In the future we could also add an input filter function so that only numbers can be entered in the number input.
Add `CLineInput::SetClipboardLineCallback` to handle the case that multiple lines of text are pasted into a lineinput. This reduces duplicate code, as this behavior was previously implemented separately for chat and console. The behavior is also fixed to be consistent with the console on Windows, so the first line being pasted edits the current input text and then sends it instead of being sent on its own without the existing input text.
Add `CalcFontSizeAndBoundingBox` to UI to reduce duplicate code. Expose `CalcAlignedCursorPos` as static member function to reuse it for line input.
Dispatch input events to UI inputs through the event handler instead of storing them in a duplicate buffer.
Use `size_t` for line input cursor position, length etc. and for `str_utf8_stats`.
Add `IButtonColorFunction` to UI to describe a functions that defines colors for the Default, Active and Hovered states of UI elements. Add some default button color functions. Use button color function to reduce duplicate code in scrollbar rendering.
Use `vec2` instead of two `floats` to represent the mouse positions in the text renderer.
Remove `CaretPosition` again, as it does not calculate the correct Y position near line breaks due to the wrapping being different when not rendering the entire string. Instead, calculate the exact caret position when rending a text container and store the caret position in the text cursor for later use.
IME usage guide (Windows)
------------------------------
1. Install the respective language and the Microsoft-IME keyboard (e.g. for Chinese, Japanese or Korean).
2. Launch the game (or a text editor to first try out the IME). Note that Windows may track the input language separately for every application. You can change this in the Windows input settings so the input language is changed globally.
2. Switch the input language using the hotkey Windows+Space or another hotkey that you configured in the Windows input settings (Alt+Shift is the default, but you should consider disabling it, to avoid accidentally changing the input language while playing).
3. Switch from Latin/English input mode to the respective asian input mode.
- Chinese: Use Ctrl+Space to switch between English and Chinese input mode. You can change this hotkey in the IME's settings.
- Japanese: Use Ctrl+Space to switch between Alphanumeric and Hiragana/Katakana input mode. You can change this hotkey in the IME's settings.
- Korean: Use Right Alt to switch between English and Hangul input mode. You cannot change this hotkey as of yet.
- Note that the input mode is also tracked per application, but there is no setting to change this behavior as far as I know, so you'll need to switch for every application separately.
4. Start typing. The underlined text is the current composition text. While a composition is active, you can only edit the composition text. Confirm the composition with Space or by selecting a candidate from the candidate list with the arrow keys. Cancel the composition with Escape or by using Backspace to delete the composition text. Note that not all languages offer a candidate list.
SDL version-specific issues
------------------------------
- 2.26.5, 2.24.2, 2.0.22: IME candidates work. But there are minor bugs when moving the composition cursor.
- 2.0.18, 2.0.20: IME candidates work.
- 2.0.16 (our current version): IME candidates cannot be determined with Windows API. Windows tries to draw the composition window like before, so this does not work in fullscreen mode.
- 2.0.8 (upstream 0.7): IME candidates work. But this SDL version is too old for us.
6268: Quit when configured bindaddr cannot be resolved, quit client when failing to open network client for 25 times r=def- a=Robyt3
## Checklist
- [X] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test (especially base/) or added coverage to integration test
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: Robert Müller <robytemueller@gmail.com>
When connecting a dummy and then logging into rcon only the dummy is logged in. When disconnecting the dummy, the main client was not automatically logged in. When logging in with the main client and then connecting the dummy, the dummy was already authenticated automatically. Now the main client is also authenticated automatically when disconnecting an authenticated dummy.
This automatic authentication was also not working correctly if the login used a username, as only the password was stored. Now both username and password are stored to correctly authenticate the main or dummy client.
The stored username and password are completely cleared when disconnecting, so they are not stored in memory longer than necessary.
Closes#5586.
Prevent endless loop when client fails to open network client forever, by quitting the client after 25 failed attempts.
As described in #5555, although it doesn't fix the actual issue.
Quit client and server if the configured bindaddr cannot be resolved.
Disable econ if configured bindaddr cannot be resolved.
To ensure that the configured bindaddr is not silently ignored.
Rename `LoadData` to `LoadDebugFont` so it's clearer what the method does.
The return value that is always `1` and not checked in some cases is removed.
The client measures the time difference between
ping send and ping reply receive.
Without MSGFLAG_FLUSH the server keeps the chunk
until the next flush which makes the ping dependend
on when the next flush happens.
6328: Remove `bytes_be_to_int` and `int_to_bytes_be`, static assert size of `int` and `unsigned`, refactoring r=heinrich5991 a=Robyt3
Supersedes #6263.
## Checklist
- [X] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test (especially base/) or added coverage to integration test
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: Robert Müller <robytemueller@gmail.com>
Use `bytes_be_to_uint` and `uint_to_bytes_be` instead.
As casting between `int` and `unsigned` preserves the bit representation of the value, it's not necessary to apply additional tricks to convert between `char` arrays and `int`.
6299: Show error message when downloaded map cannot be saved r=def- a=Robyt3
Check if deleting the old map file or renaming the temporary downloaded map fails. If so, show an error message which indicates that the user should delete the map file manually.
Sometimes downloaded map files seem to end up with wrong permissions, ownership or with read-only flag set, which makes the client unable to delete them.
Closes#5825.
## Checklist
- [X] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test (especially base/) or added coverage to integration test
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: Robert Müller <robytemueller@gmail.com>
Alternative to #6294
The only remaining problems are:
/home/deen/git/ddnet/src/engine/client/backend/glsl_shader_compiler.cpp:22:26: warning: unnecessary temporary object created while calling emplace_back [modernize-use-emplace]
m_vDefines.emplace_back(SGLSLCompilerDefine(DefineName, DefineValue));
^~~~~~~~~~~~~~~~~~~~ ~
Check if deleting the old map file or renaming the temporary downloaded map fails. If so, show an error message which indicates that the user should delete the map file manually.
Sometimes downloaded map files seem to end up with wrong permissions, ownership or with read-only flag set, which makes the client unable to delete them.
Reimplement the Linux FIFO file server and client controls on Windows by using Named Pipes.
The DDNet server/client acts as a named pipe server and receives messages.
Messages can be posted to the named pipe server by connecting to it as a client.
The named pipe client can for instance be controlled from the command line with PowerShell.
The PowerShell script `scripts/send_named_pipe.ps1` is added for this purpose.
For example the PowerShell command `./send_named_pipe.ps1 "testpipe" "echo a"` sends the command `echo a` to the pipe named `testpipe`.
Multiple commands can be sent at the same time by separating them with semicolons or newlines.
6269: Fix ddnet.org change in UUIDs r=def- a=heinrich5991
One case was a ddnet.tw UUID's string being changed (but the UUID was not), and the other case is a ddnet.tw UUID's string being changed in one place but not in another in documentation.
Fixes the commit c479230d71.
CC #5312
## Checklist
- [ ] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test (especially base/) or added coverage to integration test
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: heinrich5991 <heinrich5991@gmail.com>
One case was a ddnet.tw UUID's string being changed (but the UUID was
not), and the other case is a ddnet.tw UUID's string being changed in
one place but not in another in documentation.
Fixes the commit c479230d71.
CC #5312
6199: Register protocol and file extensions on client launch on Windows r=def- a=Robyt3
When launching the client on Windows, associate the protocol `ddnet` and the file extensions `.map` and `.demo` with the client executable.
See #6072.
## Checklist
- [X] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test (especially base/) or added coverage to integration test
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: Robert Müller <robytemueller@gmail.com>
Each time the client disconnected or stopped a demo, it tried to delete the previous temporary replay file, which causes an error message "could not delete file" to be shown in the console.
This is prevented by clearing the current filename of the demo recorder after deleting the file.
6191: Use own defines of format specifiers for MinGW (fixes#6187) r=Jupeyy a=Chairn
Counter proposition to #6188. Redefines our own specifiers
## Checklist
- [ ] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test (especially base/) or added coverage to integration test
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: Chairn <chairn.nq@hotmail.fr>
Using `IClient::LocalTime` for smooth zooming in the editor causes the zoom to behave incorrectly when joining a game, as the local time is reset when joining a game.
This is fixed by adding a separate `IClient::GlobalTime` which is only set once when the client launches and never resets.
When the client stops a replay demo and tries to remove the temporary file when being disconnected, it's not checked whether a recording of a replay demo has ever been started.
In this case the filename is empty, which leads to the client trying to delete the user's directory, which will fail with an error message, as it's a folder and cannot be deleted with the function designed for deleting files.
This is fixed by calling the function `DemoRecorder_Stop` to delete the temporary demo file, as this function already makes sure that the filename is not empty.
The variable `m_ResortServerBrowser` was only used in the client to pass it to the server browser. For better separation of concerns, this variable is moved inside `CServerBrowser` and the function `RequestResort` should be called to set it to `true`.
The existing variable `m_SortOnNextUpdate` is replaced with this, as it served the same purpose already. The variable name from upstream is used to reduce conflicts.
6035: Fix various issues reported by cppcheck static analyser r=def- a=Robyt3
After generating `compile_commands.json` with cmake, I ran [cppcheck](https://cppcheck.sourceforge.io/) like this:
```
cppcheck --project=compile_commands.json -DWIN64 --suppressions-list=cppcheck.supp --enable=all 2>cppcheck.log
```
With these suppressions in `cppcheck.supp`:
```
cstyleCast
useStlAlgorithm
unusedFunction
variableScope
noExplicitConstructor
useInitializationList
noConstructor
uninitMemberVar
uninitMemberVarPrivate
uninitDerivedMemberVar
uninitStructMember
uninitvar
shadowFunction
memleakOnRealloc
internalAstError
virtualCallInConstructor
unknownMacro
noOperatorEq
noCopyConstructor
```
Many of these occur too often or are false positives.
Here is a list of all remaining non-suppressed issues reported by cppcheck: [cppcheck.log](https://github.com/ddnet/ddnet/files/9997663/cppcheck.log)
And here is a list of all remaining issues including the suppressed ones: [cppcheck_all.log](https://github.com/ddnet/ddnet/files/9997662/cppcheck_all.log)
I couldn't get cppcheck's command line argument to ignore the external folders to work correctly, so I manually removed those entries from the files.
## Checklist
- [ ] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test (especially base/) or added coverage to integration test
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: Robert Müller <robytemueller@gmail.com>
According to cppcheck's `badBitmaskCheck` error:
```
src\engine\client\client.cpp:422:26: style: Operator '|' with one operand equal to zero is redundant. [badBitmaskCheck]
Packer.AddInt((0 << 1) | (pMsg->m_System ? 1 : 0)); // NETMSG_EX, NETMSGTYPE_EX
^
src\engine\shared\snapshot.cpp:40:45: style: Operator '|' with one operand equal to zero is redundant. [badBitmaskCheck]
int TypeItemIndex = GetItemIndex((0 << 16) | InternalType); // NETOBJTYPE_EX
^
src\engine\server\server.cpp:777:26: style: Operator '|' with one operand equal to zero is redundant. [badBitmaskCheck]
Packer.AddInt((0 << 1) | (pMsg->m_System ? 1 : 0)); // NETMSG_EX, NETMSGTYPE_EX
^
```
The client crashes when launching with `screenshot` in the command line, as the graphics are not available when the command is executed.
This is fixed by storing the command, so it's executed when everything is ready.
The types are translated to `TYPE_ALL`/`TYPE_SAVE` respectively if a given path is relative and to `TYPE_ABSOLUTE` if a path is absolute.
These types are only supported with the `OpenFile`, `ReadFile`, `ReadFileStr` and `GetCompletePath` methods.
This reduces duplicate code when calling the methods.
Fix warnings with `-fstrict-aliasing` and `-Wstrict-aliasing=2` by using char array instead of array of char pointers:
```
src/engine/client/client.cpp: In member function 'virtual const char* CClient::DemoPlayer_Play(const char*, int)':
src/engine/client/client.cpp:3858:123: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
3858 | m_aapSnapshots[g_Config.m_ClDummy][SNAP_CURRENT]->m_pSnap = (CSnapshot *)m_aaapDemorecSnapshotData[SNAP_CURRENT][0];
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^
src/engine/client/client.cpp:3859:126: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
3859 | m_aapSnapshots[g_Config.m_ClDummy][SNAP_CURRENT]->m_pAltSnap = (CSnapshot *)m_aaapDemorecSnapshotData[SNAP_CURRENT][1];
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^
src/engine/client/client.cpp:3864:117: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
3864 | m_aapSnapshots[g_Config.m_ClDummy][SNAP_PREV]->m_pSnap = (CSnapshot *)m_aaapDemorecSnapshotData[SNAP_PREV][0];
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^
src/engine/client/client.cpp:3865:120: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
3865 | m_aapSnapshots[g_Config.m_ClDummy][SNAP_PREV]->m_pAltSnap = (CSnapshot *)m_aaapDemorecSnapshotData[SNAP_PREV][1];
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^
```
Snapshot data during demo playback was being stored in an array of `char *` instead of an array of `char`, which caused above aliasing warnings and used 8 times more memory for the snapshot storage than being necessary.
6017: Minor refactoring of demo related code r=def- a=Robyt3
Extracted from #6016.
## Checklist
- [X] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test (especially base/) or added coverage to integration test
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: Robert Müller <robytemueller@gmail.com>
5599: Add support for Rust code in DDNet r=def- a=heinrich5991
The glue is done using the [cxx crate](https://cxx.rs/) on the Rust side.
As a proof-of-concept, only a small console command (`rust_version`) printing the currently used Rust version was added.
You can generate and open the Rust documentation using `DDNET_TEST_NO_LINK=1 cargo doc --open`.
You can run the Rust tests using `cmake --build <build dir> --target run_rust_tests`, they're automatically included in the `run_tests` target as well.
Rust tests don't work on Windows in debug mode on Windows because Rust cannot currently link with the debug version of the C stdlib on Windows: https://github.com/rust-lang/rust/issues/39016.
---
The stuff in `src/rust-bridge` is generated using
```
cxxbridge src/engine/shared/rust_version.rs --output src/rust-bridge/engine/shared/rust_version.cpp --output src/rust-bridge/engine/shared/rust_version.h
cxxbridge src/engine/console.rs --output src/rust-bridge/cpp/console.cpp --output src/rust-bridge/cpp/console.h
```
Co-authored-by: heinrich5991 <heinrich5991@gmail.com>
Previously the IME was not deactivated when the editor is closed, so `SDL_TEXTINPUT` events where still being reported ingame after exiting the editor.
When entering the editor with a custom bind (e.g. `bind e "cl_editor 1"`) or when opening it from the console while also immediately closing the console (e.g. with `cl_editor 1; toggle_local_console`), the IME state was not properly set to active, so SDL did not report any `SDL_TEXTINPUT` events, leading to editboxes in the editor not receiving any text.
The is fixed by always enabling the IME state when entering the editor, which was previously only done when using the Ctrl+Shift+E hotkey or coincidentally when the editor is activated while the IME is already active, i.e. when the menu or console is open.
Closes#5095.
The glue is done using the [cxx crate](https://cxx.rs/) on the Rust
side.
As a proof-of-concept, only a small console command (`rust_version`)
printing the currently used Rust version was added.
You can generate and open the Rust documentation using
`DDNET_TEST_NO_LINK=1 cargo doc --open`.
You can run the Rust tests using `cmake --build <build dir> --target
run_rust_tests`, they're automatically included in the `run_tests`
target as well.
Rust tests don't work on Windows in debug mode on Windows because Rust
cannot currently link with the debug version of the C stdlib on Windows:
https://github.com/rust-lang/rust/issues/39016.
---
The stuff in `src/rust-bridge` is generated using
```
cxxbridge src/engine/shared/rust_version.rs --output src/rust-bridge/engine/shared/rust_version.cpp --output src/rust-bridge/engine/shared/rust_version.h
cxxbridge src/engine/console.rs --output src/rust-bridge/cpp/console.cpp --output src/rust-bridge/cpp/console.h
```
5933: Inline `Is(GameType)` functions and remove support for legacy 64 player info protocol r=def- a=heinrich5991
## Checklist
- [ ] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test (especially base/) or added coverage to integration test
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
5941: Get away from vector for skins r=def- a=Jupeyy
most of the time it uses the index just to get the skin, downloaded skins change the index. Now its simply a heap object and downloaded skins load directly. Also the loading might be a bit faster bcs it had a loop lookup .Also O(1) lookup
not 100% tested. also fixes a bug with favorite skins hopefully
## Checklist
- [x] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test (especially base/) or added coverage to integration test
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: heinrich5991 <heinrich5991@gmail.com>
Co-authored-by: Jupeyy <jupjopjap@gmail.com>
It was only being used for "Leak IP" favorites and LAN servers. Since
KoG has upgraded to latest DDNet, there aren't much servers left that
don't understand the new 64 player info protocol.
Keywords: fstd, dtsf
They are used for two purposes, coloring gametypes in the serverbrowser
and enabling backward compatibility. These are independent, we shouldn't
add more stuff to the backward compatibility, hence I split them up and
inlined them.
This allows every game server to provide its own HTTPS server for map
downloads. Since the ingame protocol for downloading map data is very
inefficient, this is desirable. Previously, only servers hosted by DDNet
could benefit from this.
Security concerns:
- Attackers can find out whether a given HTTPS GET request matches a
known answer.
This isn't deemed to be problematic as no cookies for authentication
are sent and only the whole response can be matched.
- Sending requests to honeypot URLs to get people in legal trouble.
This seems to be already possible with HTML image embeds, so it can't
be that bad™.
- Downloading huge files, filling up a player's disk. The players might
cancel when seeing huge files.
There's a generous limit of 1 GiB per map file.
- Downloading huge files transparently compressed with gzip. See above.
Fixes#5812.
Add `IConsole::SetUnknownCommandCallback` to set a callback for unknown commands. The callback is used to handle connect links, .demo and .map files when parsing command line arguments.
This will allow paths/links to be passed at any argument position instead of only the first one.
And this fixes the command `play xyz.demo` not working due to `play ` being considered part of the path.
When passing .demo or .map paths as command line arguments, first check if the path can be found in the storage and then try resolving an absolute path.
Also delay execution of the `play` command until the client is ready so the command can be used from the command line. Although it doesn't work if there is only one argument, as this interferes with the .demo file handling.
Previously the playback did not start and no error messages was shown, when the map file cannot be found and no map data is embedded in the demo file.
Minor refactoring: Replace instances of `return pError` with semantically identical `return nullptr` to improve readability.
Make the member variables private and add `SetMin` to replace a usage of the member variables in `CDebugHud`. For completeness/symmetry, `SetMax` is also added.
The return value of `CGraph::InsertAt` was not checked. All uses of the function pass a correct index, so the return value is replaced with an assertion.
As the methods are always called at the same time, they can be combined. This also improves the performance, as the array only needs to be iterated once.
5658: Fix game freezing up on duplicate snapshot r=def- a=Fireball-Teeworlds
If a duplicate snapshot is received (for the same tick), we add both to the SnapshotStorage and end up with the same snapshot as both Cur and Prev. This results in GameInfraTick returning "inf" and results in "NaN" downstream in a few places, getting the CollLine logic stuck.
Some debug info (tcpdump, gdb, perf): https://gist.github.com/Fireball-Teeworlds/ad0016d2551a2e4d4cb5691023493856
Apparently this doesn't really happen in the wild, unless you have a buggy network stack. In which case it happens frequently enough to pinpoint the issue :D
(fixes#5657)
## Checklist
- [X] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test (especially base/) or added coverage to integration test
- [X] Considered possible null pointers and out of bounds array indexing
- [X] Changed no physics that affect existing maps
- [X] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: Fireball <fireball.teeworlds@gmail.com>
If a duplicate snapshot is received (for the same tick), we add both to the SnapshotStorage and end up with the same snapshot as both Cur and Prev. This results in GameInfraTick returning "inf" and results in "NaN" downstream in a few places, getting the CollLine logic stuck.
Apparently this doesn't really happen in the wild, unless you have a buggy network stack. In which case it happens frequently enough to pinpoint the issue :D
Add `CSnapshot::IsValid` to check if a snapshot unpacked from a snapshot delta or demo is valid:
- ensure number of items and data size are not negative
- ensure that the actual size of the snapshot matches the size derived from its member variables
- ensure item offsets are within the valid range
- ensure item sizes are not negative
Add `CSnapshot::TotalSize` and `CSnapshot::OffsetSize` utility functions.
Minor improvements to related error messages.
Fixes buffer overflow:
```
==47744==ERROR: AddressSanitizer: global-buffer-overflow on address 0x558618e3767f at pc 0x558614b9bdfb bp 0x7ffe58a32cd0 sp 0x7ffe58a32cc0
READ of size 4 at 0x558618e3767f thread T0
0x558614b9bdfa in CSnapshotItem::Type() const src/engine/shared/snapshot.h:16
0x558615c3c911 in CSnapshot::GetItemType(int) const src/engine/shared/snapshot.cpp:29
0x558614aebaba in CClient::UnpackAndValidateSnapshot(CSnapshot*, CSnapshot*) src/engine/client/client.cpp:2264
0x558614af87cb in CClient::OnDemoPlayerSnapshot(void*, int) src/engine/client/client.cpp:2598
0x558615b9db1a in CDemoPlayer::DoTick() src/engine/shared/demo.cpp:659
0x558615babd3f in CDemoPlayer::Update(bool) src/engine/shared/demo.cpp:1007
0x558614afb08b in CClient::Update() src/engine/client/client.cpp:2686
0x558614b1d9eb in CClient::Run() src/engine/client/client.cpp:3296
0x558614b8e64f in main src/engine/client/client.cpp:4761
```
And fixes a buffer overflow that manifests itself as an internal ASan error:
```
=================================================================
==4755==AddressSanitizer CHECK failed: ../../../../src/libsanitizer/asan/asan_descriptions.cc:79 "((0 && "Address is not in memory and not in shadow?")) != (0)" (0x0, 0x0)
0x7f0bf5f368be in AsanCheckFailed ../../../../src/libsanitizer/asan/asan_rtl.cc:72
0x7f0bf5f54eee in __sanitizer::CheckFailed(char const*, int, char const*, unsigned long long, unsigned long long) ../../../../src/libsanitizer/sanitizer_common/sanitizer_termination.cc:77
0x7f0bf5e4cb6f in GetShadowKind ../../../../src/libsanitizer/asan/asan_descriptions.cc:79
0x7f0bf5e4cb6f in __asan::GetShadowAddressInformation(unsigned long, __asan::ShadowAddressDescription*) ../../../../src/libsanitizer/asan/asan_descriptions.cc:95
0x7f0bf5e4cb6f in __asan::GetShadowAddressInformation(unsigned long, __asan::ShadowAddressDescription*) ../../../../src/libsanitizer/asan/asan_descriptions.cc:92
0x7f0bf5e4e386 in __asan::AddressDescription::AddressDescription(unsigned long, unsigned long, bool) ../../../../src/libsanitizer/asan/asan_descriptions.cc:440
0x7f0bf5e50e94 in __asan::ErrorGeneric::ErrorGeneric(unsigned int, unsigned long, unsigned long, unsigned long, unsigned long, bool, unsigned long) ../../../../src/libsanitizer/asan/asan_errors.cc:380
0x7f0bf5f35f4d in __asan::ReportGenericError(unsigned long, unsigned long, unsigned long, unsigned long, bool, unsigned long, unsigned int, bool) ../../../../src/libsanitizer/asan/asan_report.cc:460
0x7f0bf5e86f5e in __interceptor_memset ../../../../src/libsanitizer/sanitizer_common/sanitizer_common_interceptors.inc:762
0x558234873f1d in mem_zero src/base/system.cpp:213
0x55823481fc27 in CSnapshotBuilder::NewItem(int, int, int) src/engine/shared/snapshot.cpp:675
0x55823481be65 in CSnapshotDelta::UnpackDelta(CSnapshot*, CSnapshot*, void const*, int) src/engine/shared/snapshot.cpp:380
0x558234776641 in CDemoPlayer::DoTick() src/engine/shared/demo.cpp:631
0x5582347861a9 in CDemoPlayer::Update(bool) src/engine/shared/demo.cpp:1007
0x5582336d4c7d in CClient::Update() src/engine/client/client.cpp:2695
0x5582336f75dd in CClient::Run() src/engine/client/client.cpp:3305
0x558233768241 in main src/engine/client/client.cpp:4770
```
This allows a larger range of PNGs to be loaded while still maintaining
backward compatibility with older clients by annoying the user.
This warning can be enabled by the `warn-pnglite-incompatible-images`
key in the https://info2.ddnet.tw/info JSON, if the key is not there or
the JSON hasn't been obtained yet, the warning is disabled. Since the
JSON is cached across restarts, it'll be effective for initially loaded
images from the second start.
5205: Allow multiple addresses per server in the serverbrowser r=def- a=heinrich5991
Support is incomplete for `leak_ip_address_to_all_servers` (will only
ping the first address of each server) and for the `leak_ip` setting
(which will also only ping the first address of each server).
Fixes#5158.
## Checklist
- [x] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: heinrich5991 <heinrich5991@gmail.com>
This allows the client to connect to servers that have both an IPv4 and
an IPv6 address, even if the client is only connected to one of the two.
The one faster to answer will be picked.
Timeout codes are now generated based on *all* of the server's
addresses.
Fixes#5158.
Support is incomplete for `leak_ip_address_to_all_servers` (will only
ping the first address of each server) and for the `leak_ip` setting
(which will also only ping the first address of each server).
Fix pointer and pointer array variable naming
Huge renaming to match our rules
Used regex: (?!(return|delete)\b)\b\w+ (m_|ms_|g_|gs_|s_)[^a]\w+\[
(?!(return|delete)\b)\b\w+ (?!(m_|ms_|g_|gs_|s_))[^a]\w+\[
Further format static variables
Format almost all pointer names accordingly
Used regex: (?!(return)\b)\b\w+
\*(?!(m_p|p|s_p|m_ap|s_ap|g_p|g_ap|ap|gs_ap|ms_ap|gs_p|ms_p))\w+\b[^:\(p]
clang-format
Fix CI fail
Fix misnamed non pointer as pointer and non array as array
Used regex: (?!(return|delete)\b)\b\w+ (m_|ms_|g_|gs_|s_)p\w+\b
(?!return\b)\b\w+ (ms_|m_|g_|gs_|s_)a\w+\b[^\[]
clang-format
Revert to SCREAMING_SNAKE_CASE and reinstate dead code
5547: Minor tweaks to keep ODR across translation units r=def- a=Jupeyy
<!-- What is the motivation for the changes of this pull request -->
## Checklist
- [ ] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: Jupeyy <jupjopjap@gmail.com>
5486: Switch to loading screen, when map creation takes too long r=def- a=Jupeyy
Same as #4941
It doesn't directly fix the issue described in #5478, it does however not trigger it anymore. it's still a bug in our code unrelated to this. This just triggered the UB from the issue as `@ardadem` confirmed, SDL received a SDL_QUIT (`[2022-06-23 20:26:31][test]: sdl closed my client, but why xd`)
Now we have a confirmed case that making the window unresponsive can create weird behavior. Similar to the other pr switch to a loading screen after 500ms (this time without menu background map tho, since the menu background map calls the same code).
We really need some fancy loading screen for such situations, even tho they are rare xD
## Checklist
- [x] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: Jupeyy <jupjopjap@gmail.com>
Instead of closing the joysticks manually, use `SDL_QuitSubSystem(SDL_INIT_JOYSTICK)` to quit the entire subsystem, which will also close all joysticks correctly.
The engine input destructor is replaced with a `Shutdown` method so we can control when it is called, i.e. before calling `SDL_Quit`, which forcefully quits all subsystems.
`CJoystick::Close` is removed as we don't need to close joysticks manually anymore.
5429: More vector naming format r=heinrich5991 a=Chairn
Following of https://github.com/ddnet/ddnet/pull/5391
## Checklist
- [ ] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: Chairn <chairn.nq@hotmail.fr>
5406: Remove namespace tw r=Jupeyy a=heinrich5991
It didn't have a clear role, it just acted as a distinguisher between
two functions with the same name.
Rename `tw::time_get` to `time_get_nanoseconds` and delete the old
`time_get_nanoseconds`. Move `CCmdlineFix` and the typed
`net_socket_read_wait` function to the global namespace.
## Checklist
- [ ] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: heinrich5991 <heinrich5991@gmail.com>
It didn't have a clear role, it just acted as a distinguisher between
two functions with the same name.
Rename `tw::time_get` to `time_get_nanoseconds` and delete the old
`time_get_nanoseconds`. Move `CCmdlineFix` and the typed
`net_socket_read_wait` function to the global namespace.
5106: Fix dummy intended tick sent to server r=def- a=sjrc6
fixes https://github.com/ddnet/ddnet/issues/5044
Currently the dummy gives the server the wrong IntendedTick aka PredTick which (I think) makes it impossible for the server to correctly order inputs if they come early (i.e. if you're playing with high prediction margin).
Currently with dummy copy if you switch dummy with high prediction margin it will almost always desync, with this change it doesn't desync.
The only reason dummy works at all is because it relies on this code in the server to change the tick that was sent to be the current game tick, if you remove it then dummy hammer and dummy copy breaks but normal players are unaffected.
e9b59e72ff/src/engine/server/server.cpp (L1632-L1633)
I marked this PR as draft because it seems to cause some unexpected side effects:
with this change if you use dummy copy in solo part your dummy will be visually delayed, this doesn't change anything but it's different from how it was before and I don't understand why.
the sound effects from hooking with dummy copy are also not as synced as they are with the wrong tick being sent (not sure why)
I also didn't test all dummy mechanics or any advanced dummy binds which might somehow rely on the broken behavior.
With wrong tick sent to server:
https://user-images.githubusercontent.com/22122579/168413464-92a3a44b-3b75-4894-8fb0-54e64ff421fa.mp4
With correct tick:
https://user-images.githubusercontent.com/22122579/168413459-eb6f299c-260a-4ddb-9972-3dea504aef41.mp4
## Checklist
- [x] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: Tater <Mr.Potatooh@gmail.com>
5271: Time out for POST requests too (hopefully fixes#5198) r=heinrich5991 a=def-
Untested because the issue is sporadic. But I think it makes sense to have a timeout even if this is not the root cause.
## Checklist
- [ ] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: def <dennis@felsin9.de>
5064: Add HTTP masterserver registering and HTTP masterserver r=def- a=heinrich5991
Registering
-----------
The idea is that game servers push their server info to the
masterservers every 15 seconds or when the server info changes, but not
more than once per second.
The game servers do not support the old registering protocol anymore,
the backward compatibility is handled by the masterserver.
The register call is a HTTP POST to a URL like
`https://master1.ddnet.tw/ddnet/15/register` and looks like this:
```http
POST /ddnet/15/register HTTP/1.1
Address: tw-0.6+udp://connecting-address.invalid:8303
Secret: 81fa3955-6f83-4290-818d-31c0906b1118
Challenge-Secret: 81fa3955-6f83-4290-818d-31c0906b1118:tw0.6/ipv6
Info-Serial: 0
{
"max_clients": 64,
"max_players": 64,
"passworded": false,
"game_type": "TestDDraceNetwork",
"name": "My DDNet server",
"map": {
"name": "dm1",
"sha256": "0b0c481d77519c32fbe85624ef16ec0fa9991aec7367ad538bd280f28d8c26cf",
"size": 5805
},
"version": "0.6.4, 16.0.3",
"clients": []
}
```
The `Address` header declares that the server wants to register itself as
a `tw-0.6+udp` server, i.e. a server speaking a Teeworlds-0.6-compatible
protocol.
The free-form `Secret` header is used as a server identity, the server
list will be deduplicated via this secret.
The free-form `Challenge-Secret` is sent back via UDP for a port forward
check. This might have security implications as the masterserver can be
asked to send a UDP packet containing some user-controlled bytes. This
is somewhat mitigated by the fact that it can only go to an
attacker-controlled IP address.
The `Info-Serial` header is an integer field that should increase each
time the server info (in the body) changes. The masterserver uses that
field to ensure that it doesn't use old server infos.
The body is a free-form JSON object set by the game server. It should
contain certain keys in the correct form to be accepted by clients. The
body is optional if the masterserver already confirmed the reception of
the info with the given `Info-Serial`.
Not shown in this payload is the `Connless-Token` header that is used
for Teeworlds 0.7 style communication.
Also not shown is the `Challenge-Token` that should be included once the
server receives the challenge token via UDP.
The masterserver responds with a `200 OK` with a body like this:
```
{"status":"success"}
```
The `status` field can be `success` if the server was successfully
registered on the masterserver, `need_challenge` if the masterserver
wants the correct `Challenge-Token` header before the register process
is successful, `need_info` if the server sent an empty body but the
masterserver doesn't actually know the server info.
It can also be `error` if the request was malformed, only in this case
an HTTP status code except `200 OK` is sent.
Synchronization
---------------
The masterserver keeps state and outputs JSON files every second.
```json
{
"servers": [
{
"addresses": [
"tw-0.6+udp://127.0.0.1:8303",
"tw-0.6+udp://[::1]:8303"
],
"info_serial": 0,
"info": {
"max_clients": 64,
"max_players": 64,
"passworded": false,
"game_type": "TestDDraceNetwork",
"name": "My DDNet server",
"map": {
"name": "dm1",
"sha256": "0b0c481d77519c32fbe85624ef16ec0fa9991aec7367ad538bd280f28d8c26cf",
"size": 5805
},
"version": "0.6.4, 16.0.3",
"clients": []
}
}
]
}
```
`servers.json` (or configured by `--out`) is a server list that is
compatible with DDNet 15.5+ clients. It is a JSON object containing a
single key `servers` with a list of game servers. Each game server is
represented by a JSON object with an `addresses` key containing a list
of all known addresses of the server and an `info` key containing the
free-form server info sent by the game server. The free-form `info` JSON
object re-encoded by the master server and thus canonicalized and
stripped of any whitespace characters outside strings.
```json
{
"kind": "mastersrv",
"now": 1816002,
"secrets": {
"tw-0.6+udp://127.0.0.1:8303": {
"ping_time": 1811999,
"secret": "42d8f991-f2fa-46e5-a9ae-ebcc93846feb"
},
"tw-0.6+udp://[::1]:8303": {
"ping_time": 1811999,
"secret": "42d8f991-f2fa-46e5-a9ae-ebcc93846feb"
}
},
"servers": {
"42d8f991-f2fa-46e5-a9ae-ebcc93846feb": {
"info_serial": 0,
"info": {
"max_clients": 64,
"max_players": 64,
"passworded": false,
"game_type": "TestDDraceNetwork",
"name": "My DDNet server",
"map": {
"name": "dm1",
"sha256": "0b0c481d77519c32fbe85624ef16ec0fa9991aec7367ad538bd280f28d8c26cf",
"size": 5805
},
"version": "0.6.4, 16.0.3",
"clients": []
}
}
}
}
```
`--write-dump` outputs a JSON file compatible with `--read-dump-dir`,
this can be used to synchronize servers across different masterservers.
`--read-dump-dir` is also used to ingest servers from the backward
compatibility layer that pings each server for their server info using
the old protocol.
The `kind` field describe that this is `mastersrv` output and not from a
`backcompat`. This is used for prioritizing `mastersrv` information over
`backcompat` information.
The `now` field contains an integer describing the current time in
milliseconds relative an unspecified epoch that is fixed for each JSON
file. This is done instead of using the current time as the epoch for
better compression of non-changing data.
`secrets` is a map from each server address and to a JSON object
containing the last ping time (`ping_time`) in milliseconds relative to
the same epoch as before, and the server secret (`secret`) that is used
to unify server infos from different addresses of the same logical
server.
`servers` is a map from the aforementioned `secret`s to the
corresponding `info_serial` and `info`.
```json
[
"tw-0.6+udp://127.0.0.1:8303",
"tw-0.6+udp://[::1]:8303"
]
```
`--write-addresses` outputs a JSON file containing all addresses
corresponding to servers that are registered to HTTP masterservers. It
does not contain the servers that are obtained via backward
compatibility measures.
This file can be used by an old-style masterserver to also list
new-style servers without the game servers having to register there.
An implementation of this can be found at
https://github.com/heinrich5991/teeworlds/tree/mastersrv_6_backcompat
for Teeworlds 0.5/0.6 masterservers and at
https://github.com/heinrich5991/teeworlds/tree/mastersrv_7_backcompat
for Teeworlds 0.7 masterservers.
All these JSON files can be sent over the network in an efficient way
using https://github.com/heinrich5991/twmaster-collect. It establishes a
zstd-compressed TCP connection authenticated by a string token that is
sent in plain-text. It watches the specified file and transmits it every
time it changes. Due to the zstd-compression, the data sent over the
network is similar to the size of a diff.
Implementation
--------------
The masterserver implementation was done in Rust.
The current gameserver register implementation doesn't support more than
one masterserver for registering.
Co-authored-by: heinrich5991 <heinrich5991@gmail.com>
Registering
-----------
The idea is that game servers push their server info to the
masterservers every 15 seconds or when the server info changes, but not
more than once per second.
The game servers do not support the old registering protocol anymore,
the backward compatibility is handled by the masterserver.
The register call is a HTTP POST to a URL like
`https://master1.ddnet.tw/ddnet/15/register` and looks like this:
```json
POST /ddnet/15/register HTTP/1.1
Address: tw-0.6+udp://connecting-address.invalid:8303
Secret: 81fa3955-6f83-4290-818d-31c0906b1118
Challenge-Secret: 81fa3955-6f83-4290-818d-31c0906b1118:tw0.6/ipv6
Info-Serial: 0
{
"max_clients": 64,
"max_players": 64,
"passworded": false,
"game_type": "TestDDraceNetwork",
"name": "My DDNet server",
"map": {
"name": "dm1",
"sha256": "0b0c481d77519c32fbe85624ef16ec0fa9991aec7367ad538bd280f28d8c26cf",
"size": 5805
},
"version": "0.6.4, 16.0.3",
"clients": []
}
```
The `Address` header declares that the server wants to register itself as
a `tw-0.6+udp` server, i.e. a server speaking a Teeworlds-0.6-compatible
protocol.
The free-form `Secret` header is used as a server identity, the server
list will be deduplicated via this secret.
The free-form `Challenge-Secret` is sent back via UDP for a port forward
check. This might have security implications as the masterserver can be
asked to send a UDP packet containing some user-controlled bytes. This
is somewhat mitigated by the fact that it can only go to an
attacker-controlled IP address.
The `Info-Serial` header is an integer field that should increase each
time the server info (in the body) changes. The masterserver uses that
field to ensure that it doesn't use old server infos.
The body is a free-form JSON object set by the game server. It should
contain certain keys in the correct form to be accepted by clients. The
body is optional if the masterserver already confirmed the reception of
the info with the given `Info-Serial`.
Not shown in this payload is the `Connless-Token` header that is used
for Teeworlds 0.7 style communication.
Also not shown is the `Challenge-Token` that should be included once the
server receives the challenge token via UDP.
The masterserver responds with a `200 OK` with a body like this:
```
{"status":"success"}
```
The `status` field can be `success` if the server was successfully
registered on the masterserver, `need_challenge` if the masterserver
wants the correct `Challenge-Token` header before the register process
is successful, `need_info` if the server sent an empty body but the
masterserver doesn't actually know the server info.
It can also be `error` if the request was malformed, only in this case
an HTTP status code except `200 OK` is sent.
Synchronization
---------------
The masterserver keeps state and outputs JSON files every second.
```json
{
"servers": [
{
"addresses": [
"tw-0.6+udp://127.0.0.1:8303",
"tw-0.6+udp://[::1]:8303"
],
"info_serial": 0,
"info": {
"max_clients": 64,
"max_players": 64,
"passworded": false,
"game_type": "TestDDraceNetwork",
"name": "My DDNet server",
"map": {
"name": "dm1",
"sha256": "0b0c481d77519c32fbe85624ef16ec0fa9991aec7367ad538bd280f28d8c26cf",
"size": 5805
},
"version": "0.6.4, 16.0.3",
"clients": []
}
}
]
}
```
`servers.json` (or configured by `--out`) is a server list that is
compatible with DDNet 15.5+ clients. It is a JSON object containing a
single key `servers` with a list of game servers. Each game server is
represented by a JSON object with an `addresses` key containing a list
of all known addresses of the server and an `info` key containing the
free-form server info sent by the game server. The free-form `info` JSON
object re-encoded by the master server and thus canonicalized and
stripped of any whitespace characters outside strings.
```json
{
"kind": "mastersrv",
"now": 1816002,
"secrets": {
"tw-0.6+udp://127.0.0.1:8303": {
"ping_time": 1811999,
"secret": "42d8f991-f2fa-46e5-a9ae-ebcc93846feb"
},
"tw-0.6+udp://[::1]:8303": {
"ping_time": 1811999,
"secret": "42d8f991-f2fa-46e5-a9ae-ebcc93846feb"
}
},
"servers": {
"42d8f991-f2fa-46e5-a9ae-ebcc93846feb": {
"info_serial": 0,
"info": {
"max_clients": 64,
"max_players": 64,
"passworded": false,
"game_type": "TestDDraceNetwork",
"name": "My DDNet server",
"map": {
"name": "dm1",
"sha256": "0b0c481d77519c32fbe85624ef16ec0fa9991aec7367ad538bd280f28d8c26cf",
"size": 5805
},
"version": "0.6.4, 16.0.3",
"clients": []
}
}
}
}
```
`--write-dump` outputs a JSON file compatible with `--read-dump-dir`,
this can be used to synchronize servers across different masterservers.
`--read-dump-dir` is also used to ingest servers from the backward
compatibility layer that pings each server for their server info using
the old protocol.
The `kind` field describe that this is `mastersrv` output and not from a
`backcompat`. This is used for prioritizing `mastersrv` information over
`backcompat` information.
The `now` field contains an integer describing the current time in
milliseconds relative an unspecified epoch that is fixed for each JSON
file. This is done instead of using the current time as the epoch for
better compression of non-changing data.
`secrets` is a map from each server address and to a JSON object
containing the last ping time (`ping_time`) in milliseconds relative to
the same epoch as before, and the server secret (`secret`) that is used
to unify server infos from different addresses of the same logical
server.
`servers` is a map from the aforementioned `secret`s to the
corresponding `info_serial` and `info`.
```json
[
"tw-0.6+udp://127.0.0.1:8303",
"tw-0.6+udp://[::1]:8303"
]
```
`--write-addresses` outputs a JSON file containing all addresses
corresponding to servers that are registered to HTTP masterservers. It
does not contain the servers that are obtained via backward
compatibility measures.
This file can be used by an old-style masterserver to also list
new-style servers without the game servers having to register there.
An implementation of this can be found at
https://github.com/heinrich5991/teeworlds/tree/mastersrv_6_backcompat
for Teeworlds 0.5/0.6 masterservers and at
https://github.com/heinrich5991/teeworlds/tree/mastersrv_7_backcompat
for Teeworlds 0.7 masterservers.
All these JSON files can be sent over the network in an efficient way
using https://github.com/heinrich5991/twmaster-collect. It establishes a
zstd-compressed TCP connection authenticated by a string token that is
sent in plain-text. It watches the specified file and transmits it every
time it changes. Due to the zstd-compression, the data sent over the
network is similar to the size of a diff.
Implementation
--------------
The masterserver implementation was done in Rust.
The current gameserver register implementation doesn't support more than
one masterserver for registering.
These diagnostics are supposed to guide the user to problem resolution.
They're displayed if no packet is received from the server within one
second of connecting.
No message if we don't have STUN servers.
"Trying to determine UDP connectivity..." if no answer has been received
from the STUN server yet and it hasn't timed out yet.
"UDP seems to be filtered." if the STUN request has timed out.
"UDP and TCP IP addresses seem to be different. Try disabling VPN,
proxy or network accelerators." if the STUN request has returned an IP
address different from the one obtained via HTTP from info2.ddnet.tw.
"No answer from server yet." otherwise, if the STUN request has returned
no interesting data, indicating that it's likely the game server's
fault.
5085: New DDRace HUD r=def- a=C0D3D3V
If you want to test this PR, you have to test it on a Server that includes this PR too.
Textures are made by Ravie
Here a showcase video with most parts shown: https://youtu.be/gPTVj-s3pgc
Added to the new HUD
- A display of the weapons available to the player
- The weapon the player is carrying is highlighted
- Indicators for the special abilities of the player (Endless Hook, Endless Jumps, Jetpack, Teleport Weapons)
- Indicators for abilities taken away from the player (Deep/Life Freeze, No Hook, No Weapons collision, No Collision)
- Control indicators for dummy controls (dummy hammer, dummy copy) (bottom right)
- Jump indicator (max 10 jumps ar displayed, and greyed out as soon as a jump is used)
- Ninja status bar that indicates how long a player is capable of using ninja (next to the ninja sword)
- Freeze status bar that indicates the thawing time of a player (below player)
- Movement Information can be displayed in a clean way above the mini score HUD (Position, Speed, Target Angle)
- Indicator if you are in practice mode
The complete HUD also works for players you spectate
I Added a new NetObj since the predicted values are not perfect and would make the display of the information a lot less good: DDNetCharacterDisplayInfo that contains the following information
```
NetIntRange("m_JumpedTotal", -2, 255),
NetTick("m_NinjaActivationTick"),
NetTick("m_FreezeTick"),
NetBool("m_IsInFreeze"),
NetBool("m_IsInPracticeMode"),
NetIntAny("m_TargetX"), # used for the Movement Information display
NetIntAny("m_TargetY"),
NetIntAny("m_RampValue"),
```
So if someone has an idea what data we could also need in the client for making the display more nice, now is the right moment to add more data to this network object.
A few screenshots:
Assets Tab:
HUD Settings:
Mini Debug HUD:
Speed.X is calculated using the players ramp vaule
## Checklist
- [x] Tested the change ingame
- [x] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [x] Considered possible null pointers and out of bounds array indexing
- [x] Changed no physics that affect existing maps
- [x] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: c0d3d3v <c0d3d3v@mag-keinen-spam.de>
Co-authored-by: Jupeyy <jupjopjap@gmail.com>
* master: (87 commits)
Remove base/tl/string.h
Replace remaining usage of base/tl/string with std::string
Remove unused includes of base/tl/string.h
Store localized strings in a CHeap instead of using tl/string.h
Mark methods as const
Add CHeap::StoreString method
Rules are chat responses too
Add margins to demo slice popup, decrease error font size, UI scaling
Remove redundant parameters which are overridden later
Use Margin instead of both VMargin and HMargin
Move variable declaration
Only output messages intended for chat to the user of a chat command
Remove unused chat response variables
Don't print the first "Waiting for score threads to complete"
fix usage of undefined behavior for default eyes
remove duplicate HOOK_RETRACTED assignment
do not send swap request notification to complete team 0
make swap messages more personal
Move ninja shield to other position (fixes#5047)
do not release the hooks if you swap
...
This makes the "black console window" less important on Windows (or
anywhere else, for that matter), lets you see logs from other threads in
the f1 console, and removes the distinction between `IConsole::Print`
and `dbg_msg`.
4509: Switch new players to DDNet config dir, support Teeworlds as fallback r=heinrich5991 a=def-
This way new players will get DDNet directory, old ones can switch
directory if they want, or keep using the old one.
If we ever enforce a switch in a future version, this will make it
easier since older DDNet versions will also support the DDNet directory
already.
<!-- What is the motivation for the changes of this pull request -->
## Checklist
- [x] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: def <dennis@felsin9.de>
Co-authored-by: Dennis Felsing <dennis@felsin9.de>
This way new players will get DDNet directory, old ones can switch
directory if they want, or keep using the old one.
If we ever enforce a switch in a future version, this will make it
easier since older DDNet versions will also support the DDNet directory
already.
- Added RenderMovementInformation() to the HUD
- Use m_Width instead of recalculating the display width each time in hud.cpp
- Reworked Debug HUD:
- Display velspeed.x*ramp and velspeed.y instead of velspeed*ramp
- Show unit of speed
- The graph for the velspeed.x*ramp behavior (if dbg_tuning is active) is now nicer and usable. A second graph zooming in on the turning point has been added.
- The velspeed.x*ramp graph will be redrawn only when new tuning parameters are added
- The target angle used to draw the tees is now calculated using the m_TargetX, m_TargetY in DDNetCharacterDisplayInfo if available.
- Added m_TargetX, m_TargetY and m_RampValue to DDNetCharacterDisplayInfo Snap (m_RampValue has a precision of one thousandth)
- Added CGraph::InsertAt() to create static Graphs
- Added "assets/hud" to initial created Directories
- Added Settings to hide Dummy Action HUD and Player Movement HUD cl_showhud_dummy_actions, cl_showhud_player_position, cl_showhud_player_speed, cl_showhud_player_angle
- Added Pointer to m_PrevExtendedDisplayInfo, for calculation of interpolated Angle in Player Rendering
- Added GetDigitsIndex(int Value, float Max); to hud.cpp
> Warn when a local variable shadows another local variable or parameter.
Found one actual bug in graphics_threaded.cpp
Should reduce confusion in the future when reading source code
4783: Some network refactoring r=def- a=heinrich5991
In preparation to the proxy code, I refactored the network code a little, unifying its interface across operating systems and making our use of recvmmsg invisible to the caller.
## Checklist
- [x] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: heinrich5991 <heinrich5991@gmail.com>
4758: Fix logging into rcon with dummy r=def- a=heinrich5991
Fixes#4721.
<!-- What is the motivation for the changes of this pull request -->
## Checklist
- [x] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: heinrich5991 <heinrich5991@gmail.com>
src/engine/client/client.cpp:3861:47: error: ‘INT_MAX’ was not declared in this scope
3861 | if(Start < 0 || Length < 0 || Start > INT_MAX - Length)
| ^~~~~~~
src/engine/client/client.cpp:81:1: note: ‘INT_MAX’ is defined in header ‘<climits>’; did you forget to ‘#include <climits>’?
4628: Disable or fix some spammy new clang-tidy 13 warnings r=heinrich5991 a=def-
```
src/engine/client/backend/opengles/opengles_sl_program.cpp:10:11: warning: suspicious #include of file with '.cpp' extension [bugprone-suspicious-include]
^
src/engine/graphics.h:330:67: warning: 2 adjacent parameters of 'CFreeformItem' of similar type ('float') are easily swapped by mistake [bugprone-easily-swappable-parameters]
CFreeformItem(float x0, float y0, float x1, float y1, float x2, float y2, float x3, float y3) :
^~~~~~~~~~~~~~~~~~
src/game/editor/layer_tiles.cpp:1802:14: warning: result of multiplication in type 'int' is used as a pointer offset after an implicit widening conversion to type 'ptrdiff_t' [bugprone-implicit-widening-of-multiplication-result]
mem_zero(&m_pSwitchTile[y * m_Width], o * sizeof(CSwitchTile));
^
src/game/server/entities/plasma.cpp💯152: warning: operator has equivalent nested operands [misc-redundant-expression]
if(SnapPlayer && SnapPlayer->GetTeam() != TEAM_SPECTATORS && !SnapPlayer->IsPaused() && SnapChar && SnapChar && SnapChar->Team() != m_ResponsibleTeam && SnapPlayer->m_ShowOthers != 1)
~~~~~~~~ ~~~~~~~~ ^
src/game/editor/editor.cpp:4295:38: warning: The right operand of '>' is a garbage value [clang-analyzer-core.UndefinedBinaryOperatorResult]
if(m_FilePreviewImageInfo.m_Width > Preview.w)
^ ~~~~~~~~~
```
<!-- What is the motivation for the changes of this pull request -->
## Checklist
- [ ] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
4639: Fix SendStartWarning() to only warn every 3 seconds r=heinrich5991 a=def-
Thanks for Skeith for report
<!-- What is the motivation for the changes of this pull request -->
## Checklist
- [x] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: def <dennis@felsin9.de>
Co-authored-by: Dennis Felsing <dennis@felsin9.de>
On the one hand variables called "Dummy" would tell us whether the
current action refers to the currently inactive tee ("dummy"). On the
other hand, these variables could tell us whether the current action
refers to the main connection to the server, or the secondary one. The
latter use case is now renamed to "Client", with the choices
`CLIENT_MAIN`, `CLIENT_DUMMY` (and `CLIENT_CONTACT`).
Perhaps better names could be found, especially since `Client` also
refers to the engine client class in the game code.
I tried to not fix bugs unless it would complicate the code.
4581: Adjustable prediction margin r=def- a=trml
This adds an option to adjust how early the client sends an input for a given tick. In vanilla this value is hardcoded to 10 milliseconds before the next tick, which is good for low-latency gameplay and/or when you have a stable connection, but sometimes it could perhaps be useful to increase the value, to either test something with a high ping (without having to joing a high-ping server or use on a 3rd party program like tc/netem), or for counteracting the effect of jitter and jumping ping.
The last part comes at a tradeoff for higher ping though, and it also doesn't handle lag caused by packet loss, but it I have found it useful in some situations when playing race/solo/dummy with a bad connection. Since it's a bit experimental I only added it as an f1 option and not stored between sessions, and also added a gameinfo flag to only support it on servers that implement #1441.
Edit: Also attempted to cleanup variable names for tick calculations in client.cpp slightly.
## Checklist
- [x] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [x] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [ ] Considered possible null pointers and out of bounds array indexing
- [x] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
4590: Remove extra projectiles r=def- a=trml
Aligns the code with vanilla, and cleans up things a bit
(same as [329e6261f3](url))
These were used to show projectiles a tick (or fraction of a tick) earlier than they otherwise would (and they were also sent without extrainfo since they didn't matter for prediction), so removing them would perhaps not make a noticeable difference. (and perhaps also less of a difference now since weapon input isn't applied before the start of a tick anymore).
## Checklist
- [x] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
4593: Fix compiler warning with new ffmpeg version (currently built from git) r=heinrich5991 a=def-
```
src/engine/client/video.cpp:102:32: error: assigning to 'AVOutputFormat *' from 'const struct AVOutputFormat *' discards qualifiers
m_pFormat = m_pFormatContext->oformat;
~~~~~~~~~~~~~~~~~~^~~~~~~
src/engine/client/video.cpp:571:13: error: assigning to 'AVCodec *' from 'const AVCodec *' discards qualifiers
*ppCodec = avcodec_find_encoder(CodecId);
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
```
<!-- What is the motivation for the changes of this pull request -->
## Checklist
- [ ] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [ ] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [ ] Considered possible null pointers and out of bounds array indexing
- [ ] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
4607: Adopt upstream input (especially double click) handling, refactoring r=def- a=Robyt3
This adopts the 0.7 way of handling double clicks, by checking `Event.button.clicks` of the `SDL_Event`. The double click is stored as a flag which is cleared when the `MouseDoubleClick()` function is called (d215c73206 and 7977b46d36).
This reverts 72a6e20, as this hack is no longer required (closes#1745). Double clicking a friend in the friend list still works (#444) with both `gfx_asyncrender_old 0` and `1`.
Refactorings and minor fixes synchronizing with upstream:
- Only copy SDL keystate of _keyboard_ keys and mem_zero the rest to ensure out of bounds SDL keys don't trigger mouse keys.
- Rename variables and use SDL constants.
- Reduce unnecessary indentation.
## Checklist
- [X] Tested the change ingame
- [ ] Provided screenshots if it is a visual change
- [X] Tested in combination with possibly related configuration options
- [ ] Written a unit test if it works standalone, system.c especially
- [X] Considered possible null pointers and out of bounds array indexing
- [X] Changed no physics that affect existing maps
- [ ] Tested the change with [ASan+UBSan or valgrind's memcheck](https://github.com/ddnet/ddnet/#using-addresssanitizer--undefinedbehavioursanitizer-or-valgrinds-memcheck) (optional)
Co-authored-by: trml <trml@users.noreply.github.com>
Co-authored-by: Dennis Felsing <dennis@felsin9.de>
Co-authored-by: Robert Müller <robert.mueller@uni-siegen.de>
Co-authored-by: oy <Tom_Adams@web.de>
otherwise they get stuck. Thanks to Skeith reproduction steps:
> Basically, to consistently reproduce the bug, you put your tee in
> freeze, then you connect your dummy right as the stars appear when
> frozen. It also happens with the stars that pop out when you shoot your
> pistol @deen
ChillerDragon
