5252: Use std::vector instead of array, remove base/tl/array.h, algorithm.h, allocator.h, range.h r=def- a=Robyt3
Replaces all usages of `array` with `std::vector`.
I adjusted variable names of variables I changed to use the `v` prefix. Not so in the editor however, as there are already many changes due to clang-tidy enforcing the use of for-each loops.
This allows us to remove all remaining `base/tl` headers except `threading.h`.
Clang-tidy now finds `clang-analyzer-cplusplus.NewDelete` (Use of memory after it is freed), which is also fixed here, though it appears to be a false-positive.
This last remaining usages of `goto` are also removed.
## 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: Robert Müller <robytemueller@gmail.com>
5132: Refactoring: Use std::swap in editor r=heinrich5991 a=Robyt3
<!-- 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: Robert Müller <robytemueller@gmail.com>
4687: Add Vulkan backend r=def- a=Jupeyy
This adds a Vulkan 1.0 backend and new features to the graphic settings to select the GPU(if multiple Vulkan GPUs are supported, and a new list for renderers (see screenshot below))
Mutleasy benchmark (CPU bound):
Vulkan multithreaded, single threaded vs OpenGL
(lower graphs are histograms, code by `@Chairn` )
Y = Frametimes in microseconds
lower graphs = amount of frametimes that happened (histogram)
(since the renderer speed differs, you should look at it more like a spread of values rather than the actual values)
Vulkan is especially good in these scenarios, beating OpenGL 3.3 almost 3x with my setup (~600-700 FPS vs. ~1700-1800FPS)
Remaining TODO list:
- [x] compile shaders in cmake ( e.g. `https://gist.github.com/evilactually/a0d191701cb48f157b05be7f74d79396` )
- [ ] needs windows vulkan libraries
- [x] add build instructions (packages)
- [x] get away from coherent memory even for staging buffers (flushing memory just seems to be faster)
- [ ] a lot of testing :P
If there are bugs and it's crashing inside the driver the best you can do is to start the client with `dbg_gfx 4` which will (if supported) add Khronos standard validation layers + verbose debugging information + validation layer extensions, e.g. a synchronization validation layer and validation errors reported by the GPU driver directly.
(setting for dbg_gfx (0: none, 1: minimal, 2: affects performance, 3: verbose, 4: all))
edits:
fixes#3547 (probably fixes it, fixed some data races)
## 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
- [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: Jupeyy <jupjopjap@gmail.com>
Co-authored-by: Dennis Felsing <dennis@felsin9.de>
> 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
4768: Videorecorder on by default (also for github builds) r=Jupeyy a=def-
<!-- 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: Dennis Felsing <dennis@felsin9.de>
Since it causes crashes because another thread just accesses the data.
No one has wanted to fix this, so maybe we should just remove it instead
of having clients crash in editor. I'm wondering how much effort it
would be to make this safe. Would we just have to lock for a short part
or is it basically the entire threaded undo operation that has to be
serialized? (No need for a thread then)
ChillerDrgon
