- Show Player/Dummy tabs at the top instead of using checkbox for dummy settings like in regular skin settings.
- Show Basic/Custom tabs instead of using button to toggle custom skin settings.
- Move Skin directory and Refresh icon buttons to right side like in regular skin settings. Refreshing 0.7 skins and parts is not implemented yet though.
- Render 0.7 skin entries like entries of the regular skin list, i.e. with more space and with only four skins per row.
- Move the Random skin button next to the Custom colors checkbox and make it an icon button like in the regular skin settings.
- Move skin preview to the right side of the Player/Dummy/Basic/Custom tabs to reduce unused empty space.
- Remove most overlapping and unnecessarily dark backgrounds.
- Improve layout of skin part tabs by only using colors for the first and last buttons.
- Localize skin part tab names again but properly capitalize the names being shown in the UI.
Reduce duplicate and inconsistent code for rendering quick search for the demo browser, ingame vote list, player flag, skin, skin 0.7 and asset search.
The quick search and exclude in the server browser are not refactored, as they have additional labels and different alignment, which would make a general function complicated.
Avoid Vulkan crash if the backend is destroyed immediately after being created.
Slightly decreases time of initial black screen before loading menu is rendered.
Convert mapres to RGBA immediately when loading them, so the image data is always in RGBA format internally, instead of only converting when the map is saved (which was erroneously removed in #8670).
This means the `cl_editor_dilate` setting will now also be applied to converted RGB images.
The macro `__ARM_ARCH` is defined both for 32-bit and 64-bit ARM so it cannot be used to identify ARM64. Now `__ARM_ARCH_ISA_A64` is used instead, which should only be defined for ARM64. This caused a warning due to the macro `CONF_ARCH_STRING` being redefined when compiling for Android. Furthermore, support for detecting big-endian ARM64 with the `__ARM_BIG_ENDIAN` macro is added.
See https://developer.arm.com/documentation/dui0774/g/chr1383660321827
Add a button with the "terminal" icon in the bottom right of the start menu to open the local console to ensure that the local console is usable also when no physical keyboard (with F-keys) is available.
Do not reset the active map download's information before using the fallback map download.
Remove redundant calls of `ResetMapDownload` before disconnecting, as this already resets the map download.
Closes#8885. Regression from #8848.
If the game is paused and a player joins a server (sv_tournament_mode 0)
The scoreboard will be forced open. Unless the client configured cl_scoreboard_on_death 0.
This can be quite annoying. Especially in the brand new 0.7 feature
where users can pause the game. Oy realized that this is a problem 12
year ago:
aec468a3c4 (diff-e0ff7a1d6079610adb64fc89fbfff23a381ed92f268d8fe188731a9e0c323b0aR389-R390)
For ddnet servers this would mostly affect tournaments where paused games are
used to give everyone enough time to download the map.
A open scoreboard also blocks broadcasts. So new users might miss the
admin announcements explaining why the game is paused.
Most but not all sound creations in CCharacter::FireWeapon have this NOLINT comment.
Somehow clang does not get tripped on all of those. But sooner or later it might.
I currently got a new error in my downstream fork when tweaking
FireWeapon a bit. So I applied the NOLINT comment to ALL occurences in
FireWeapon not only the ones that clangd finds at the moment.
Following sequence of events was possible:
1. HTTP map download for map X is started (e.g. client initially connects to server).
2. Map change message for map Y arrives while downloading (e.g. vote for another map has passed on the server).
3. Client already has map Y and loads it directly without starting or resetting the map download.
4. HTTP map download task for map X finishes.
5. Map X is loaded, overriding the map Y that should be loaded. This is one cause for #2804.
This could potentially also happen with the map download from the game-server, if the final map data chunk for a downloaded map is received directly after another map change message for a map which does not need to be downloaded, though this seems more unlikely.
This is fixed by resetting the current map download immediately after receiving a valid map change message instead of only when starting a new map download.
The relevant code to reset the map download is moved to `CClient::ResetMapDownload` to avoid duplicate code and ensure that the temporary map download file is always deleted and closed.
* Fixes OpenGL on Wayland when the GLX flavor of GLEW is linked
* Only if SDL2's video driver is "wayland" we also allow the
GLEW_ERROR_NO_GLX_DISPLAY error code to be accepted as if glewInit()
actually succeeded
* Hopefully this won't be needed anymore on future versions of GLEW when
glewContextInit() is made part of the public API
5cf0e5e997/datasrc/network.py (L8)5cf0e5e997/datasrc/seven/network.py (L10)
0.6 game flags:
```python
GameFlags = ["TEAMS", "FLAGS"]
```
0.7 game flags:
```python
GameFlags = Flags("GAMEFLAG", ["TEAMS", "FLAGS", "SURVIVAL", "RACE"])
```
The 0.7 protocol extended the game flags without changing the old 0.6 flags.
So flag teams and flag flags are the same in both versions and 0.7 just
has additional flags.
The server is now sending the same m_GameFlags from the gamemode
controller to 0.6 and 0.7 connections. It clamps away the unsupported
flags for 0.6.
Before this commit the ddnet server would always send the hardcodet race
flag to 0.7 connections while ignoring what the gamemode set as a game
flag.
Support using emoticon selector and spectator menu with touch inputs. Touching in the emoticon selector activates the selected (eye) emote and closes the emoticon selector. The spectator menu is kept open when using it for more convenience. The emoticon selector and spectator menu can be closed by touching anywhere outside of them.
Additionally, the emoticon selector and specator menu can now be closed by pressing the Escape key (i.e. the back button on Android). This made it necessary to change the order of the components for input handling, so the ingame menu will handle the Escape key after the emoticon selector and spectator menu. This also means that the Escape key can now be used to close the selector/menu when they are stuck open, which was previously delayed until ingame menu or console were opened and closed.
The additional margin left of the Spectate and Join game buttons is unnecessary.
Avoid reusing `s_SpectateButton` variable for different buttons.
Use `TEAM_RED` instead of `0` when joining game in non-team server.
This reverts commit 91848f0be6.
This was prone to accidental usage. The user interface is now usable with touch input, so this additional method to quickly and cleanly quit the game is not necessary anymore.
Use the same HSLA scrollbar color picker for the 0.7 tee settings as for the 0.6 tee settings. Use the correct darkest lighting value for 0.7 skin color pickers instead of not clamping the lighting value in the color picker.
Use existing color constructor to conditionally unpack the alpha component instead of doing this separately.
Make `DARKEST_COLOR_LGT` a `float` constant instead of using an `enum` to simplify the usage.
Rename `CMenus::RenderHSLScrollbars` function to `RenderHslaScrollbars`.
Pass darkest light value as `float` to the `RenderHslaScrollbars` function to add support for different values, instead of enabling the clamping with a `bool` parameter. Previously, this parameter was always set to `true`, as this type of color picker is only used for skins.
Remove unnecessarily complicated code for rendering unclamped lighting scrollbar. The code can be simplified by considering the darkest lighting value as a variable.
Let the `RenderHslaScrollbars` function return `true` if the color was changed to simplify the usage. The function previously returned the color but this value was unused.
Use `IGraphics::SetColor4` function instead of `IGraphics::SetColorVertex` when possible. Rename variables `Array` to `aColorVertices`. Use `std::size` instead of hard-coding array sizes.
On macOS, semaphore names are global and processes can open semaphores of other processes by using their name. It was possible that multiple client/server processes randomly tried to create semaphores with the same memory address, which would cause them to have the same name. In that case, `sem_open` would fail because `O_EXCL` prevents creating semaphores if the name is already used.
This is made more unlikely by also including the PID in the semaphore name. Additionally, the semaphore name is prefixed with `/` to more closely follow the construction rules for semaphore names. In particular, the behavior is implementation-defined if the name is not prefixed with a slash.
See https://pubs.opengroup.org/onlinepubs/9799919799/functions/sem_open.htmlCloses#8773.
More details are added to the assertion error message in case this does not end up fixing the issue.
Players are supposed to turn of their VPN or complain to the admins if
they get banned for using a VPN. Calling this "banned" seems to be
confusing, as players ask for an unban even after they turned off the
VPN.
Simplify clipboard handling by returning an `std::string` and freeing the string returned by SDL immediately, so the clipboard data does not stay in memory unnecessarily after the clipboard has been used until the clipboard data is requested again.
Fix possible TOCTOU when pasting from the clipboard into a lineinput, due to the clipboard data being requested twice.
- Fix value `0` being used for sound volume, so the tee sound was not audible.
- Use `DoButtonLogic` to trigger sound to ensure it does not play while other UI elements are active.
- Remove unnecessay `MousePosition` variable by using `MousePos` function instead.
Update src/engine/shared/netban.cpp
Remove the word "currently" for the empty ban list message.
Co-authored-by: Dennis Felsing <dennis@felsing.org>
Update src/engine/shared/netban.cpp
Text format refactor, adding "currently"
Co-authored-by: Dennis Felsing <dennis@felsing.org>
Mark remaining `CCollision` functions as `const` when possible.
Add remaining getter functions for all tile data and mark all tile getter functions as `const`.
Avoid duplicate calculation by extraction result of `GetPureMapIndex` in variable.
Rename variable `Indices` to `Index`.
Include `vector` instead of `list`, as the former is used but the latter is not.
Also show client IDs in the spectator menu when `cl_show_ids` is enabled. Previously, this setting applied to scoreboard and chat, although the setting and config variable descriptions were only mentioning the scoreboard.
Extract client ID formatting in `CGameClient::FormatClientId` function to reduce duplicate code and ensure it is consistent.
Correctly indent client IDs also when largest client ID is 100 or larger. Fix scoreboard spectator client ID buffer not being large enough for IDs 100 and larger.
Make client ID formatting more efficient by avoiding `str_format` except for formatting the number itself and appending text directly to the text cursor when possible.
Previously, the `net_udp_create` function returned whether the IPv6 socket in particular was created successfully. If an IPv4 socket or Websocket was created successfully before this, it was not used and also not closed properly, if the IPv6 socket could not be created.
Now, the `net_udp_create` function succeeds if the socket for at least one network type (IPv4, IPv6 or Websocket) could be created successfully.
Avoid reusing local variables and remove redundant comments. Also apply the same refactoring to `net_tcp_create`, which already has the behavior of succeeding if the socket for any network type (IPv4 or IPv6) was created successfully.
We already sent the legacy is ddnet message and so far the ddnet
protocol over 0.7 seems to work fine.
But for completeness and correctness it is now also sending the new
ddnet version net message.
- Only log successful log messages when `debug` is enabled, consistent with logging for regular skins.
- Use `IsImageFormatRgba` to check for RGBA image format, so this also shows a warning popup.
- Remove redundant logic for non-RGBA image data.
- Use `CheckImageDivisibility` instead of enforcing exact size of xmas hat and bot decoration images to allow higher resolution images.
- Fix image data not being freed if it has the wrong format or ratio.
- Fix potential integer overflow when calculating skin part blood color (same as for regular skins).
- Extract `LoadXmasHat` and `LoadBotDecoration` functions for readability.
skins7: Extract the skins dir to a macro
skins7: Look for the skins at 'skins7' dir
Data: Add 0.7 skins (diliated)
CMenus::RenderSkinSelection7: Fix the skins number (use 0.7 getter)
CSkins7::GetColorV3: Fix missing UnclampLighting
CRenderTools: Use stronger type for 0.7 colors
CSkins7::RandomizeSkin: Use enum value instead of hardcode+comment
CMenus::RenderSettingsTee7: Enable team skins validation
system: Add str_utf8_copy_num()
GameClient: Extract ApplySkin7InfoFromGameMsg()
GameClient: Add skin7 to CClientData (also process and validate it)
While keeping 0.6 fully working and untouched
This adds the option to connect via 0.7
0.7 servers are now mixed into the server browser list.
And users can also manually connect via 0.7
using a connection url like so:
./DDNet "connect tw-0.7+udp://127.0.0.1:8303"
This also adds a "Tee 0.7" section in the menu to pick a 0.7 skin.
The config variables are prefixed with player7_ like for example
player7_color_body
- Remove `readability-identifier-naming.StaticVariableIgnoredRegexp`.
- Mark `static` variables as `const` and use `UPPER_CASE` for names when possible.
- Use local variables instead of `static` variables when possible.
- Use member variables with correct name instead of `static` variables.
- Pass output buffer for `static` function `GetKeyBindModifiersName` by argument instead of returning pointer to `static` buffer.
Previously, binding `+` commands like `+hook` to composite keys like `shift+x` or `ctrl+alt+mouse1` was not possible, as it would lead to stuck inputs when one of the modifiers is released first before the primary key. When binding `+` commands to composite binds the modifier was instead ignored to prevent this.
Now, binding `+` commands to composite keys is also possible. Active binds (key, modifier) are stored and when a modifier key is released all binds using that modifier are also released.
Closes#8314.
The handling of F1-F24 keys in `CBindsSpecial` was not correct for composite binds, leading to stuck inputs if modifier keys are released first. Now, the `CBindsSpecial::OnInput` function simply delegates to `CBinds::OnInput` for F-key presses/releases so the behavior is consistent with regular binds.
- Use `UnbindAll` function in destructor to reduce duplicate code.
- Use appropriate constants instead of magic number `0`.
- Remove redundant comments.
- Remove redundant `else if` condition.
- Rename generic loop variables `i` and `j` to `Modifier` and `Key` for clarity.
This was only affecting the key names displayed in the voting HUD, e.g. with `bind "ctrl+alt+f3" "vote yes"` the voting HUD previously showed `+f3` as the key name.
The `OnInput` function should return `true` exactly for events that were handled. In case of line inputs, the function returned `true` when then line input was changed instead. For the consoles, the function did not return `true` for some events that were handled.
The flag `TEXLOAD_NO_COMPRESSION` is entirely unused.
The flag `TEXLOAD_NOMIPMAPS` is unused and only the respective `CCommandBuffer::TEXFLAG_NOMIPMAPS` is used internally, so the former flag does not need to be exposed to users of `IGraphics`.
The `IGraphics::LoadTextureRawSub` function and the respective `CMD_TEXTURE_UPDATE` command are unused. Only text textures are updated after having been uploaded, but there is a separate `CMD_TEXT_TEXTURE_UPDATE` for that.
It should not be necessary to access the null-texture directly. The `CTextureHandle::IsNullTexture` function can be used to check if a texture handle is the null-texture.
Assume that this branch should apply to all team modes except `SV_TEAM_FORCED_SOLO` also when new team modes would get added after `SV_TEAM_FORCED_SOLO` in the future, as the values will never be changed and there is no relation between the order of the numeric values and the behavior.
Previously, swapping on forced solo servers was already impossible, as this would require two players to join the same team, so it would always fail with the error `Player is on a different team` (or earlier). The check `g_Config.m_SvTeam != 3` was never reached. Also, this check was not using the constant `SV_TEAM_FORCED_SOLO`. Now, an earlier check is added to show a more specific error message when trying to use `/swap` on a forced solo server.
The additional indirection with the `CEffects::ResetDamageIndicator` and `CDamageInd::Reset` functions to reset damage indicators is confusing and unnecessary. Always reset the damage indicators when disconnecting (in `OnReset`) instead of when entering the next game.
Rename `IGraphics::IsImageFormatRGBA` function to `IsImageFormatRgba`.
Rename parameter `pFileName` to `pContextName`, as this name does not necessarily describe a file.
Remove unnecessary check for parameter `pFileName` (now `pContextName`) being unset, which is and should never be the case.
Rename parameter `Img` to `Image` for consistency.
Put the context name in quotation marks in the warning message.
Rename parameter `pFileName` to `pContextName`, as this name does not necessarily describe a file.
Rename parameter `Img` to `Image` for consistency.
The function is unused and because the message is not flushed it would also not have worked correctly. The `NETMSG_INPUT` message is already packed and flushed in the `CClient::SendInput` function.
Closes#8408.
The maximum width and height were swapped as the loop bounds, so the menu checker background did not have the correct width on very wide resolutions.
This was not noticed on common resolutions because twice as many quads as necessary were also being rendered for each row of the background. The bounds of the inner loop are adjusted and only every second x-value is iterated, which reduced the total number of quads for the checker background rendering by more than half.
Track whether the dummy is currently connecting separately so the dummy being disconnected due to errors can be detected. Show an error message as echo in the chat when the dummy could not be connected, e.g. when the server is full.
Use the global time for limiting the dummy connecting delay instead of using game ticks, so the delay also works correctly when having connection problems (i.e., when ticks do not advance). Handle the dummy connecting being delayed separately from the dummy currently connecting.
Add tooltips for the "Connect dummy"-button when it's disable due to dummy not being allow on the server or when connecting is delayed.
Add console error messages for `dummy_connect` command.
The image data was not being freed when `IGraphics::LoadTextureRawMove` is used with images that are not in RGBA format as this falls back to using `IGraphics::LoadTextureRaw` which requires manually freeing the image data.
Also consistently clear all image info in `LoadTextureRawMove`. This makes it necessary to store the size of the preview image in the editor separately, as previously the width and height of the unloaded image info were being used to render the preview image.
Previously, IME composition strings were limited to at most 32 bytes, e.g. 10 Hiragana characters. Now, `SDL_TextEditingExtEvent` (`SDL_TEXTEDITING_EXT`) is supported, which uses heap-allocated composition strings of arbitrary length.
On the other hand, the buffer for input event text was larger than necessary, as SDL text events only contain at most 32 bytes (`SDL_TEXTINPUTEVENT_TEXT_SIZE` is not used directly to avoid the include). With the hint for extended IME handling enabled, long composition text will be sent as multiple text events (without breaking UTF-8).
Add constants for BPP and alpha threshold during Dilate, as dilating only works for RGBA images and we only use a fixed alpha threshold.
Move local variable declarations closer to their usages.
Use `mem_copy` instead of for-loop.
Use `nullptr` instead of `NULL`.
ChillerDragon
