There's now only 3 addressing modes: stack, constant, and member.
Self, class, and nil are now present respectively in the first 3 stack
slots. Global and class constants are moved to local constants when
compiling. Named globals is only present on editor to use on tool
singletons, so its use now emits a new instruction to copy the global to
the stack.
This allow us to further optimize the VM later by embedding the
addressing modes in the instructions themselves, which is better done
with less permutations.
This is meant for testing the GDScript implementation, not for testing
user scripts nor testing the engine using scripts.
Tests consists in a GDScript file and a .out file with the expected
output. The .out file format is: expected status (based on the enum
GDScriptTest::TestStatus) on the first line, followed by either an error
message or the resulting output. Warnings are added after the first
line, before the output (or compiler errors) if the parser pass without
any error.
The test script must have a function called `test()` which takes no
argument. Such function will be called by the test runner. The test
should not have any dependency unless it's part of the test too. Global
classes (using `class_name`) are registered before the runner starts, so
those should work if needed.
Use the command `godot --gdscript-generate-tests
godot-source/modules/gdscript/tests/scripts` to update the .out files
with the current output (make sure the output are the expected values
before committing).
The tests themselves are part of the doctest suite so those can be
executed with `godot --test`.
Co-authored-by: Andrii Doroshenko (Xrayez) <xrayez@gmail.com>
When the type cannot be validated at compile time, the runtime must do a
check to ensure type safety is kept, as the code might be assuming the
return type is correct in another place, leading to crashes if the
contract is broken.
This ensures that annotations that rely on the datatype (such as
@export) can validated it timely, allowing compound expressions instead
of only literal values.
- Use `Array[type]` for type-hints. e.g.:
`var array: Array[int] = [1, 2, 3]`
- Array literals are typed if their storage is typed (variable
asssignment of as argument in function all). Otherwise they are
untyped.
When this code was changed for 4.0, a "break" statement inside a for loop in 3.x was changed to "return".
This means that the two special cases (autoloads and input actions) are never checked.
Removing the return lets these work properly in the editor.
(Also reorder conditionals to short-circuit and avoid expensive methods.)
-Added a new method in Resource: reset_state , used for reloading the same resource from disk
-Added a new cache mode "replace" in ResourceLoader, which reuses existing loaded sub-resources but resets their data from disk (or replaces them if they chaged type)
-Because the correct sub-resource paths are always loaded now, this fixes bugs with subresource folding or subresource ordering when saving.
- Based on C++11's `thread` and `thread_local`
- No more need to allocate-deallocate or check for null
- No pointer anymore, just a member variable
- Platform-specific implementations no longer needed (except for the few cases of non-portable functions)
- Simpler for `NO_THREADS`
- Thread ids are now the same across platforms (main is 1; others follow)
Happy new year to the wonderful Godot community!
2020 has been a tough year for most of us personally, but a good year for
Godot development nonetheless with a huge amount of work done towards Godot
4.0 and great improvements backported to the long-lived 3.2 branch.
We've had close to 400 contributors to engine code this year, authoring near
7,000 commit! (And that's only for the `master` branch and for the engine code,
there's a lot more when counting docs, demos and other first-party repos.)
Here's to a great year 2021 for all Godot users 🎆
Instead of references. This is needed because those function pointers
are used in GDNative which needs to work with plain C, which doesn't
support passing parameters by reference.
Storing script references to pointer only in result.script_type could
lead to losing the last reference, causing further conversions from
Script* to Ref<Script> to fail.
Now result.script_type_ref is always set first, and then cleared in the
specific case of the script being the owner, to avoid cyclic reference
issues.
The removed check was adding a protection for the case where a `Reference` has not yet got its reference count initialized and a script is called on it. That would cause the object to be released after the call. The removed code was constructing the `Variant` via the `Object` constructor so it didn't deal with the reference count and so the release was prevented.
However, `Variant` no longer works that way so that check was useless. Now it's just illegal to run GDScript on a Reference whose reference count has not been initialized.
- Initialize Object pointer to nullptr so it's not used by mistake.
- When setting an Object check if it's a reference so refcounting works
as intended.
Values that are passed by reference are not suited for being constructed
at compile time because in this case they would be shared across all the
construction statements.
- ClassDoc added to GDScript and property reflection data were extracted
from parse tree
- GDScript comments are collected from tokenizer for documentation and
applied to the ClassDoc by the GDScript compiler
- private docs were excluded (name with underscore prefix and doesn't
have any doc comments)
- default values (of non exported vars), arguments are extraced from the
parser
- Integrated with GDScript 2.0 and new enums were added.
- merge conflicts fixed
- Use the new functions in Variant to determine the validity and resulting
type of operators.
- Split the operator function in codegen between binary and unary, since
the unary ones have now a special requirement of having the second
argument to be the NIL type when requesting info.
They are now called "utility functions" to avoid confusion with methods
of builtin types, and be consistent with the naming in Variant.
Core utility functions are now available in GDScript. The ones missing
in core are added specifically to GDScript as helpers for convenience.
Some functions were remove when there are better ways to do, reducing
redundancy and cleaning up the global scope.
The call of range() in a for loop is optimized to use int or vectors, to
avoid allocating an array, however the type was set as array still. With
the new typed VM this is an issue as the type mismatch the actual value,
resulting in wrong instructions to be selected.
- Replace the for loop temporaries by locals. They cause conflicts with
the stack when being popped, while locals are properly handled in the
scope.
- Change the interface for the codegen so the for loop list doesn't live
through the whole block if it's a temporary.
- Keep track of the actual amount of local variables in the stack. Using
the size of the map is misleading in cases where multiple locals have
the same name (which is allowed when there's no shadowing).
- Added a few debug checks for temporaries, to avoid them being wrongly
manipulated in the future. They should not live more than a line of
code.
- Rearrange some of compiler code to make sure the temporaries don't
live across blocks.
Almost all instructions need variant arguments. With this change they
are loaded in an array before each instruction call. This makes the
addressing code be localized to less places, improving compilation
overhead and binary size by a small margin.
This should not affect performance.
* Using C-style function pointers now, InternalMethod is gone.
* This ensures much better performance in typed code.
* Renamed builtin_funcs to utility_funcs, to avoid naming confusion
The underscore prefix was used to avoid the conflict between the `RID` class
name and the matching enum value in `Variant::Type`.
This can be fixed differently by prefixing uses of the `RID` class in `Variant`
with the scope resolution operator, as done already for `AABB`.
-Discern between named, indexed and keyed
-Get direct access to functions for typed GDScript and GDNative bindings
-Small changes to some classes in order to work with the new setget binder
The GDScript `load` mention is moved from the class `ResourceLoader`
description to the `ResourceLoader.load` method description instead,
where it is more likely to be found.
Fixes#41462 where calling Object.new() in GDScript gave an error.
I fixed it by adding exclusion when checking if the name is a builtin
type to exclude objects with a comment detailing why.
This ensures that scripts created without a resource loader are properly
included in the cache (such as builtin scripts) and are not tried to be
loaded from the disk.
The input to smoothstep is not actually a weight, and the decscription
of smoothstep was pretty hard to understand and easy to misinterpret.
Clarified what it means to be approximately equal.
nearest_po2 does not do what the descriptions says it does. For one,
it returns the same power if the input is a power of 2. Second, it
returns 0 if the input is negative or 0, while the smallest possible
integral power of 2 actually is 1 (2^0 = 1). Due to the implementation
and how it is used in a lot of places, it does not seem wise to change
such a core function however, and I decided it is better to alter the
description of the built-in.
Added a few examples/clarifications/edge-cases.
In general they are more confusing to users because they expect
inheritance to fully override parent methods. This behavior can be
enabled by script writers using a simple super() call.
Sometimes to fix something you have to break it first.
This get GDScript mostly working with the new tokenizer and parser but
a lot of things isn't working yet. It compiles and it's usable, and that
should be enough for now.
Don't worry: other huge commits will come after this.
My initial attempt changed this in the gdscript code, which resulted in
a duplicate warning name in the builtin editor. We should just append
the warning name in the LSP instead.
This uses parens to match what is shown in the builtin editor.
- Extacted all syntax highlighting code from text edit
- Removed enable syntax highlighting from text edit
- Added line_edited_from signal to text_edit
- Renamed get/set_syntax_highlighting to get/set_syntax_highlighter
- Added EditorSyntaxHighligher
Occasionally you want to ignore a warning with a `warning-ignore`
comment, and you have to go into the settings to look up what the
actual name of the warning is. This patch appends the warning name to
the end of the warning so you know what string to use to ignore it,
similar to other linters like pylint.
For example
```
"The signal 'blah' is declared but never emitted.";
```
is now
```
"The signal 'blah' is declared but never emitted. (UNUSED_SIGNAL)";
```
Reverts `latest_client_id` back to 0, as I misunderstood how the client
IDs are assigned and, without further testing and debugging, I can't
say if this was a bug or a valid default value.
Similarly, a `latest_client_id` of -1 is no longer raising an error.
Fixes#39548.
`latest_client_id` now defaults to `-1` (invalid ID) instead of `0`.
Also fix typo in notification `gdscrip_client/changeWorkspace`,
and fix argument names in method binds.
Fixes#39375.
Depending on the conditional statements of the 'for' and 'while' loops,
their body may not even execute once. For example:
func a():
var arr = []
for i in arr:
return i
# can be reached, but analysis says cannot
return -1
func b():
var should_loop = false
while should_loop:
return 1
# can be reached, but analysis says cannot
return 0
The parser will complain that the statements after the comment cannot
be reached, but it is clearly possible for our scenario. This is
because the parser falsely assumes that the loop body will always
execute at least once.
Fix the code to remove this assumption for both of those loops.
I couldn't find a tool that enforces it, so I went the manual route:
```
find -name "thirdparty" -prune \
-o -name "*.cpp" -o -name "*.h" -o -name "*.m" -o -name "*.mm" \
-o -name "*.glsl" > files
perl -0777 -pi -e 's/\n}\n([^#])/\n}\n\n\1/g' $(cat files)
misc/scripts/fix_style.sh -c
```
This adds a newline after all `}` on the first column, unless they
are followed by `#` (typically `#endif`). This leads to having lots
of places with two lines between function/class definitions, but
clang-format then fixes it as we enforce max one line of separation.
This doesn't fix potential occurrences of function definitions which
are indented (e.g. for a helper class defined in a .cpp), but it's
better than nothing. Also can't be made to run easily on CI/hooks so
we'll have to be careful with new code.
Part of #33027.
Which means that reduz' beloved style which we all became used to
will now be changed automatically to remove the first empty line.
This makes us lean closer to 1TBS (the one true brace style) instead
of hybridating it with some Allman-inspired spacing.
There's still the case of braces around single-statement blocks that
needs to be addressed (but clang-format can't help with that, but
clang-tidy may if we agree about it).
Part of #33027.
Using `clang-tidy`'s `modernize-use-default-member-init` check and
manual review of the changes, and some extra manual changes that
`clang-tidy` failed to do.
Also went manually through all of `core` to find occurrences that
`clang-tidy` couldn't handle, especially all initializations done
in a constructor without using initializer lists.
Part of #33027, also discussed in #29848.
Enforcing the use of brackets even on single line statements would be
preferred, but `clang-format` doesn't have this functionality yet.
Now the stack saved in a `GDScriptFunctionState` is cleared as soon as the `yield()` operation is known not to be resumed because either the script, the instance or both are deleted.
This clears problems like leaked objects by eliminating cases of circular references between `GDScriptFunctionState`s preventing them and the objects they refer to in their saved stacks from being released. As an example, this makes using `SceneTreeTimer` safer.
Furthermore, with this change it's now possible to print early warnings about `yield()`s to released script/instances, as now we know they won't be successfully resumed as the condition for that happens. However, this PR doesn't add such messages, to keep the observed behavior the same for the time being.
Also, now a backup of the function name in `GDScriptFunctionState` is used, since the script may not be valid by the time the function name is needed for the resume-after-yield error messages.
This is achieved by skipping initializer call while creating an instance
of a GDScript. This is implemented by passing -1 as an argument count
to `_new` and interpreting any value below 0 to mean that the initializer
should not be called during instantiation, because internal members of
an instance are going to be overridden afterwards.
Calling _reduce_node_type from GDScriptParser::_parse_block for assert
was using a current class with a scope that didn't include all
functions. Now calling in GDScriptParser::_check_block_types uses the
right class type. We also now check the assert node message. The assert
line was added to the set_errors associated with assert, since before
the error would be reported on the next line
Note: Only replaced 2 instances to test, Node.get_children and TileMap.get_used_cells
Note: Will do a mass replace on later PRs of whathever I can find, but probably need
a tool to grep through doc.
Warning: Mono will break, needs to be fixed (and so do TypeScript and NativeScript, need to ask respective maintainers)
Also added an easier way to load native GLSL shaders.
Extras:
Had to fix no-cache for subresources in resource loader, it was not properly working, making shaders not properly reload.
Note:
The precommit hooks are broken because they don't seem to support enums from one class being used in another.
Feel free to fix this after merging this PR.
Also implemented decal atlas, so projectors and other stuff can be added.
Sidenote: Had to make RID hashable, so some unrelated includes changed
in order to include it in hashfuncs.h
Configured for a max line length of 120 characters.
psf/black is very opinionated and purposely doesn't leave much room for
configuration. The output is mostly OK so that should be fine for us,
but some things worth noting:
- Manually wrapped strings will be reflowed, so by using a line length
of 120 for the sake of preserving readability for our long command
calls, it also means that some manually wrapped strings are back on
the same line and should be manually merged again.
- Code generators using string concatenation extensively look awful,
since black puts each operand on a single line. We need to refactor
these generators to use more pythonic string formatting, for which
many options are available (`%`, `format` or f-strings).
- CI checks and a pre-commit hook will be added to ensure that future
buildsystem changes are well-formatted.
Now calling _reduce_node_type with debugging enabled to determine
if assert line is safe. Part of doing this required the assert line
to be stored away. Now the AssertNode line is being correctly set.
Newlines are now marked safe always
Now that the unused DocDump was removed, the `editor/doc` subfolder is
redundant.
Similarly, there's no reason for Collada to have a subfolder for itself
when glTF or OBJ don't.
For us, it practically only changes the fact that `A<A<int>>` is now
used instead of the C++03 compatible `A<A<int> >`.
Note: clang-format 10+ changed the `Standard` arguments to fully
specified `c++11`, `c++14`, etc. versions, but we can't use `c++17`
now if we want to preserve compatibility with clang-format 8 and 9.
`Cpp11` is still supported as deprecated alias for `Latest`.
EngineDebugger is the new interface to access the debugger.
It tries to be as agnostic as possible on the data that various
subsystems can expose.
It allows 2 types of interactions:
- Profilers:
A subsystem can register a profiler, assigning it a unique name.
That name can be used to activate the profiler or add data to it.
The registered profiler can be composed of up to 3 functions:
- Toggle: called when the profiler is activated/deactivated.
- Add: called whenever data is added to the debugger
(via `EngineDebugger::profiler_add_frame_data`)
- Tick: called every frame (during idle), receives frame times.
- Captures: (Only relevant in remote debugger for now)
A subsystem can register a capture, assigning it a unique name.
When receiving a message, the remote debugger will check if it starts
with `[prefix]:` and call the associated capture with name `prefix`.
Port MultiplayerAPI, Servers, Scripts, Visual, Performance to the new
profiler system.
Port SceneDebugger and RemoteDebugger to the new capture system.
The LocalDebugger also uses the new profiler system for scripts
profiling.
Main:
- It's now implemented thanks to `<mutex>`. No more platform-specific implementations.
- `BinaryMutex` (non-recursive) is added, as an alternative for special cases.
- Doesn't need allocation/deallocation anymore. It can live in the stack and be part of other classes.
- Because of that, it's methods are now `const` and the inner mutex is `mutable` so it can be easily used in `const` contexts.
- A no-op implementation is provided if `NO_THREADS` is defined. No more need to add `#ifdef NO_THREADS` just for this.
- `MutexLock` now takes a reference. At this point the cases of null `Mutex`es are rare. If you ever need that, just don't use `MutexLock`.
- Thread-safe utilities are therefore simpler now.
Misc.:
- `ScopedMutexLock` is dropped and replaced by `MutexLock`, because they were pretty much the same.
- Every case of lock, do-something, unlock is replaced by `MutexLock` (complex cases where it's not straightfoward are kept as as explicit lock and unlock).
- `ShaderRD` contained an `std::mutex`, which has been replaced by `Mutex`.
- Renames PackedIntArray to PackedInt32Array.
- Renames PackedFloatArray to PackedFloat32Array.
- Adds PackedInt64Array and PackedFloat64Array.
- Renames Variant::REAL to Variant::FLOAT for consistency.
Packed arrays are for storing large amount of data and creating stuff like
meshes, buffers. textures, etc. Forcing them to be 64 is a huge waste of
memory. That said, many users requested the ability to have 64 bits packed
arrays for their games, so this is just an optional added type.
For Variant, the float datatype is always 64 bits, and exposed as `float`.
We still have `real_t` which is the datatype that can change from 32 to 64
bits depending on a compile flag (not entirely working right now, but that's
the idea). It affects math related datatypes and code only.
Neither Variant nor PackedArray make use of real_t, which is only intended
for math precision, so the term is removed from there to keep only float.
This attribute is now part of the standard we target so we no longer
need compiler-specific hacks.
Also enables -Wimplicit-fallthrough for Clang now that we can properly
support it. It's already on by default for GCC's -Wextra.
Fixes new warnings raised by Clang's -Wimplicit-fallthrough.
WARNING: Requires C++17 'guaranteed copy elision' to fix ambiguous
operator problems in Variant.
This was added for this commit (and future C++17 uses) in #36457.
This is needed because of the new changes to Variant. The reference
counter is increased by adding it to a Variant, which means no GDScript
will be freed (or will be double freed if manually freed somewhere).
