When `IMap::Load` fails, other components will continue to use the old map. However, if `IMap::Load` failed after the map was already successfully read with the datafile reader then other components kept their pointers to the old, invalid datafile reader items and data, causing random crashes in collision code. This is fixed by using a separate datafile reader to read the new map and only applying the datafile reader globally when loading was entirely successfully.
An error message is added for the case that a map has an unsupported version, which is currently the only case where a map can fail to load after the datafile was read successfully.
In particular, the block maps `blmapPepe`, `blmapV5` and `blmapDT-UPTU` did not have a version map item and were fixed separately.
Closes#7218. Regression from #5737.
Instead of keeping track of a permanently empty `CSnapshot` object in client and server separately, add `CSnapshot::EmptySnapshot` to access a singleton empty `CSnapshot`.
Mark pointer parameters of snapshot functions as `const` when possible.
Previously, if the demo header strings did not contain zero-termination, the client would render the strings and any following non-zero memory from the demo header.
Now, demos will not be loaded, if any string in the header is not zero-terminated or not valid UTF-8.
Most variables used in the sound engine were static globals, as they are used in the static sound mixing function. The global variables are replaced by member variables, by passing the sound interface as user-data for the SDL mixing callback. The `Mix` function is made a public member function of `ISound` instead of being exposed using `ISoundMixFunc GetSoundMixFunc()`.
This allows to remove the direct dependency of the engine sound on the engine video, by instead passing the sound mixing function as a lambda to the engine video in the engine client.
The old WavPack reader function interface does support passing a user-data pointer to the callback function, so global variables are still used here.
Don't truncate console lines at 255 bytes anymore. Especially lines containing many Unicode characters would be adversely affected by this limitation.
Instead, truncate console lines after 10 wrapped lines are rendered. Rendering too many lines at once currently breaks the console scrolling. Rendering an ellipsis is currently not possible when rendering text with a maximum line count.
Increase buffer sizes to handle long (esp. invalid) command inputs.
Closes#7132.
Use `WaitForPipeDrain` to deterministically wait for the pipe to drain instead of using `Start-Sleep`.
Use `Dispose` instead of `Close` to properly flush and close the pipe stream.
Add error handling for connection timeout and I/O errors.
Handle `ERROR_BROKEN_PIPE` separately when peeking at pipe, as this happens when the pipe is disconnected immediately after connecting it or after reading the previous message.
Don't ignore `ERROR_BAD_PIPE` anymore, as the pipe should never be in a disconnected (i.e. bad) state at this point of the function.
This config variable was only settable in the server console but only read in the client, so it was effectively unusable. It also has no use case right now.
Render a pause/play icon in the center of the screen when pausing/unpausing while the menu is not active.
The icon fades in and out over 0.5 seconds and slightly increases in size over that time.
When starting demo rendering with initial pause, the pause indicator is rendered continuous until playback is first started, to ensure that the initial pause state is communicated clearly to the user. The initial pause indicator is not included in the rendered demo, but pausing and unpausing later during demo rendering will cause it to be included in the video, same as other UI elements. Closes#7044.
The config variable `cl_demo_show_pause` (`0/1`, default `1`) is added to hide the pause indicator entirely, for example to render a demo with multiple pauses without the indicator.
The pause indicator and also the existing speed indicator are not rendered (anymore) while the menu is active, as the menu already contains this information.
Add more efficient function for formatting integer values as strings.
A benchmark shows that using this function is significantly faster than using `str_format`. It is faster by a factor of 220 with Clang 15.0 O2 (https://quick-bench.com/q/BlNoLnlyqxipf4jvsFTUxKMHDJU) and by a factor of 11 with GCC 12.2 O2 (https://quick-bench.com/q/Fxf9lDCTqXBF4pIa_IyZ5R0IqYg).
This increases FPS in the editor by ~25% when many numbers are rendered for switch/tele/speedup/tune layers or with "Show Info" being enabled.
The additional static analysis for `std::to_chars` revealed that the wrong size was used in `CHud` for `aScoreTeam[TEAM_RED]` and `aScoreTeam[TEAM_BLUE]`.
This requires incrementing the macOS deployment target from 10.13 to 10.15.
`shared_ptr`s of `CServer::m_pDnsblLookup[ClientID]` are set, but not
cleaned up when done. Therefore the Job is kept alive until the player
disconnects and a new player joins on that slot. Currently this means
that the full linked list of jobs is kept alive.
When the Job is overwritten with a new job, all the remaining objects in
the list can be dropped. With enough jobs, that is causing a stack
overflow in the destructor.
This patch fixes this overflow by making the lifetime independent of the
previous Job. Jobs can get dropped after processing them, even when the
previous job is still alive.
Fixes#6954
The old code looked pretty weird. It checked for `CFGFLAG_SERVER |
CFGFLAG_GAME` being set above, and then used the passed-in parameter to
find the command. I think it'd be better to just look for `CFGFLAG_GAME`
and then find the command using `CFGFLAG_GAME`.
Check if datafile data cannot be read entirely (according to the data size specified in the header) and check for decompression errors. In case of errors, let `GetData` return `nullptr` and `GetDataSize` return `0` for the respective index.
Internally the decompressed size is set to `-1` for data which failed to load, so loading of those data will not be attempted again because it would only fail again and can cause additional log messages.
Port the `CJsonWriter` utility class from upstream, which makes outputting correct JSON easier.
Add `CJsonWriter` as an abstract class that can write to different outputs. Two implementations `CJsonFileWriter` (writes to a file) and `CJsonStringWriter` (writes to an `std::string`) are added. Upstream `CJsonWriter` can only write to files.
The same tests are added for both implementations. Duplicate code is avoided by using typed tests with two separate test fixtures.