* GDScript call stack as reverse linked list with issues fixed
(originally proposed in 91006).
* Fix coroutine issues with call stack by resuming async call chain
inside `GDScriptFunction::call()`.
* This fixes corrupted line numbers for coroutines in the debugger and
backtrace (106489).
Co-authored-by: Juan Linietsky <reduzio@gmail.com>
When the types are validated at compile time, this type of call runs
faster. It is already used for instance methods, this adds this
optimization to native static methods as well.
Fixes the issue by adding a mechanism by which the functions that were
previously disappearing can be profiled too. This is optional with
an editor setting, since collecting more information naturally slows the engine
further while profiling.
Fixes#23715, #40251, #29049
This improves the performance of typed calls to engine methods when the
argument types are exact.
Using validated calls delegate more of the work the core instead of
doing argument unpacking in the VM. It also does not need different
instructions for each return type, simplifying the code.
This assumes that operators are called usually with the same type of
operands as the first time. So it stores the types of the first run and
if matched it uses an optimized path by calling the validated operator
function directly. Otherwise it uses the regular untyped evaluator.
With this change, if operators do use the same type they run quite
faster. OTOH, if the types mismatch it takes longer to run than they
would with the previous code.
* This implementation adds threads on the side of the client (script debugger).
* Some functions of the debugger are optimized.
* The profile is also now thread safe using atomics.
* The editor can switch between multiple threads when debugging.
This PR adds threaded support for the script language debugger. Every thread has its own thread local data and it will connect to the debugger using multiple thread IDs.
This means that, now, the editor can receive multiple threads entering debug mode at the same time.
This applies our existing style guide, and adds a new rule to that style
guide for modular components such as platform ports and modules:
Includes from the platform port or module ("local" includes) should be listed
first in their own block using relative paths, before Godot's "core" includes
which use "absolute" (project folder relative) paths, and finally thirdparty
includes.
Includes in `#ifdef`s come after their relevant section, i.e. the overall
structure is:
- Local includes
* Conditional local includes
- Core includes
* Conditional core includes
- Thirdparty includes
* Conditional thirdparty includes
Which allows editable data associated with a particular class instead of
the instance. Scripts with static variables are kept in memory
indefinitely unless the `@static_unload` annotation is used or the
`static_unload()` method is called on the GDScript.
If the custom function `_static_init()` exists it will be called when
the class is loaded, after the static variables are set.
The hard limit is set at 2048 depth which seems sensible between
legitimate recursive calls while still avoiding a crash because of a
stack overflow in most of the cases.
Note that it is still possible to reach the stack limit and get an
overflow before reaching a call depth. This is intended as a half-way
measure to stop crashing in most cases, since there's no reliable nor
portable way to check the amount of stack memory left.
When instructions use function pointers, it's not possible to retrieve
their original names in the disassembly. This stores the names in
vectors (in debug builds) so they can be shown.
As many open source projects have started doing it, we're removing the
current year from the copyright notice, so that we don't need to bump
it every year.
It seems like only the first year of publication is technically
relevant for copyright notices, and even that seems to be something
that many companies stopped listing altogether (in a version controlled
codebase, the commits are a much better source of date of publication
than a hardcoded copyright statement).
We also now list Godot Engine contributors first as we're collectively
the current maintainers of the project, and we clarify that the
"exclusive" copyright of the co-founders covers the timespan before
opensourcing (their further contributions are included as part of Godot
Engine contributors).
Also fixed "cf." Frenchism - it's meant as "refer to / see".
* Removed instruction argument count and instruction prefetching. This is now done on the fly. Reduces jumps.
* OPCODE_DISPATCH now goes directly to the next instruction, like in Godot 3.x.
I have nothing I can use to test performance, so if anyone wants to lend a hand and compare with master (both on debug and release), it would be very welcome.
- RPC configurations are now dictionaries.
- Script.get_rpc_methods renamed to Script.get_rpc_config.
- Node.rpc[_id] and Callable.rpc now return an Error.
- Refactor MultiplayerAPI to allow extension.
- New MultiplayerAPI.rpc method with Array argument (for scripts).
- Move the default MultiplayerAPI implementation to a module.
Implement built-in classes Vector4, Vector4i and Projection.
* Two versions of Vector4 (float and integer).
* A Projection class, which is a 4x4 matrix specialized in projection types.
These types have been requested for a long time, but given they were very corner case they were not added before.
Because in Godot 4, reimplementing parts of the rendering engine is now possible, access to these types (heavily used by the rendering code) becomes a necessity.
**Q**: Why Projection and not Matrix4?
**A**: Godot does not use Matrix2, Matrix3, Matrix4x3, etc. naming convention because, within the engine, these types always have a *purpose*. As such, Godot names them: Transform2D, Transform3D or Basis. In this case, this 4x4 matrix is _always_ used as a _Projection_, hence the naming.
When a type is shared (i.e. passed by reference) it doesn't need to be
called in a setter chain (e.g. `a.b.c = 0`) since it will be updated in
place.
This commit adds an instruction that jumps when the value is shared so
it can be used to skip those cases and avoid redundant calls of setters.
It also solves issues when assigning to sub-properties of read-only
properties.
* Map is unnecessary and inefficient in almost every case.
* Replaced by the new HashMap.
* Renamed Map to RBMap and Set to RBSet for cases that still make sense
(order matters) but use is discouraged.
There were very few cases where replacing by HashMap was undesired because
keeping the key order was intended.
I tried to keep those (as RBMap) as much as possible, but might have missed
some. Review appreciated!
We prefer to prevent using chained assignment (`T a = b = c = T();`) as this
can lead to confusing code and subtle bugs.
According to https://en.wikipedia.org/wiki/Assignment_operator_(C%2B%2B), C++
allows any arbitrary return type, so this is standard compliant.
This could be re-assessed if/when we have an actual need for a behavior more
akin to that of the C++ STL, for now this PR simply changes a handful of
cases which were inconsistent with the rest of the codebase (`void` return
type was already the most common case prior to this commit).
