Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Removed faulty function update after get_property_list. #63712

Merged
merged 1 commit into from
Aug 6, 2022

Conversation

object71
Copy link
Contributor

@object71 object71 commented Jul 31, 2022

The function tried to rearrange properties but that lead to problems with duplication or deleted properties. Implemented the logic that that function did inside the get_property_list both for tool scripts and non-tool scripts.

Fixes #63668

There shouldn't be an update function that rearranges properties in the first place. If there was a requirement for properties to be split into categories based on the class inheritance this should've initially been implemented into GDScript or GDScriptInstance function for getting properties.

NOT TESTED on GDExtension and C# scripts but it should work.

@object71 object71 requested review from a team as code owners July 31, 2022 08:10
@object71 object71 force-pushed the fix-export-issues branch from 8bc06ad to 0dfffe9 Compare July 31, 2022 08:17
@object71 object71 force-pushed the fix-export-issues branch 5 times, most recently from f87320d to a677eae Compare July 31, 2022 12:15
@object71
Copy link
Contributor Author

I cannot compile for mono, can anybody confirm that c# scripts look okay innthe inspector both for rool an non tool scripts with inheritance?

@raulsntos
Copy link
Member

raulsntos commented Jul 31, 2022

To compile mono you need to rebase and apply PR #63537 because currently the mono module is broken in master (Edit: the PR was merged and the mono module should be working in master now). With those changes master should compile, but your PR still has an issue, it can be fixed with this patch:

diff --git a/modules/mono/csharp_script.cpp b/modules/mono/csharp_script.cpp
index 0e3c17557c..62463e787a 100644
--- a/modules/mono/csharp_script.cpp
+++ b/modules/mono/csharp_script.cpp
@@ -1806,7 +1806,6 @@ void CSharpInstance::get_property_list(List<PropertyInfo> *p_properties) const {
 	}
 
 #ifdef TOOLS_ENABLED
-	Ref<Script> script = get_script();
 	if (script->is_valid()) {
 		props.push_front(script->get_class_category());
 	}

This is probably what you intended to do, you were shadowing the script member with get_script which just returns it with a less specific type so we can't access the script_class and native members later. With those changes this PR seems to be working fine.

@object71
Copy link
Contributor Author

object71 commented Aug 1, 2022

To compile mono you need to rebase and apply PR #63537 because currently the mono module is broken in master (Edit: the PR was merged and the mono module should be working in master now). With those changes master should compile, but your PR still has an issue, it can be fixed with this patch:

diff --git a/modules/mono/csharp_script.cpp b/modules/mono/csharp_script.cpp
index 0e3c17557c..62463e787a 100644
--- a/modules/mono/csharp_script.cpp
+++ b/modules/mono/csharp_script.cpp
@@ -1806,7 +1806,6 @@ void CSharpInstance::get_property_list(List<PropertyInfo> *p_properties) const {
 	}
 
 #ifdef TOOLS_ENABLED
-	Ref<Script> script = get_script();
 	if (script->is_valid()) {
 		props.push_front(script->get_class_category());
 	}

This is probably what you intended to do, you were shadowing the script member with get_script which just returns it with a less specific type so we can't access the script_class and native members later. With those changes this PR seems to be working fine.

Thanks, will fix that one!

@object71 object71 force-pushed the fix-export-issues branch 4 times, most recently from 1838a40 to 92675e9 Compare August 1, 2022 06:52
@object71
Copy link
Contributor Author

object71 commented Aug 1, 2022

Managed to get mono running and fixed the inspector for mono too. It now looks good both for inherited, tool and non-tool scripts. :)

@object71 object71 force-pushed the fix-export-issues branch from 92675e9 to 0ad894e Compare August 1, 2022 07:33
@raulsntos
Copy link
Member

Awesome, thank you for adding categories to mono 👍

It seems the changes from push_front to push_back are causing the exported properties in mono to be reversed now:

GDScript C#
gd csharp

@object71
Copy link
Contributor Author

object71 commented Aug 1, 2022

Ah, ok. Will fix the ordering.

@object71
Copy link
Contributor Author

object71 commented Aug 3, 2022

Reveresed the properties addition for C#.

@lufog
Copy link
Contributor

lufog commented Aug 3, 2022

Checked on the artifact from actions, @export_category(...), @export_group(...) and @export_subgroup(...) do not work.

More

Code:

@export_category("Category 1")
@export_group("Group A", "group_a")
@export var group_a_var_a: float
@export var group_a_var_b: float
@export_group("Group B", "group_b")
@export var group_b_var_a: float
@export var group_b_var_b: float
@export_group("Group C", "group_c")
@export_subgroup("Subgroup A", "group_c_a")
@export var group_c_a_var_a: float
@export var group_c_a_var_b: float
@export_subgroup("Subgroup B", "group_c_b")
@export var group_c_b_var_a: float
@export var group_c_b_var_b: float

Expected:
image

Current:
image

@object71
Copy link
Contributor Author

object71 commented Aug 3, 2022

Thanks for the note. I forgot to test those new features. Will see to it.

@YuriSizov
Copy link
Contributor

YuriSizov commented Aug 3, 2022

Checked on the artifact from actions, @export_category(...), @export_group(...) and @export_subgroup(...) do not work.

They seem to only work in tool scripts now. Which is very weird, because they are parsed together with other annotations.

Double-checked, and it works correctly in alpha 13 and current master.

Edit: I guess in non-tool scripts script-defined categories and groups are stripped for some reason? And _update_script_class_properties re-introduced them by happenstance? That would also explain why it was unnoticed for the old-school way to define groups with _get_property_list, as that requires a tool script to begin with.

@YuriSizov
Copy link
Contributor

YuriSizov commented Aug 3, 2022

@object71 I figured how to fix the grouping annotations properly, you need to add a case in GDScript::_update_exports() to include them in member cache.

Something like this seems to be enough:

	for (int i = 0; i < c->members.size(); i++) {
		const GDScriptParser::ClassNode::Member &member = c->members[i];

		switch (member.type) {
			// ...

			case GDScriptParser::ClassNode::Member::GROUP: {
				members_cache.push_back(member.annotation->export_info);
			} break;
			default:
				break; // Nothing.
		}
	}

Unlike regular variable members they don't have default values, but this doesn't seem to crash anything. Worst case scenario, some usages of member_default_values_cache may need to be guarded.

@YuriSizov
Copy link
Contributor

I've tested some chaotic stuff with the aforementioned fix, and it seems to work as expected. For example, defining a category at the top of your script would hide the base script category, effectively overriding it. Pretty sweet!

godot.windows.tools.64_2022-08-03_20-26-40.mp4

There is a bug, though — when you change the exports on the base class, the inspector for the extending class does not react to these changes. You need to reload the scene to see the changes properly. This is probably outside of the scope for this particular PR, however.

Code looks way better now, and IMO makes sense. I'd like a review from @vnen on this.

modules/gdscript/gdscript.cpp Outdated Show resolved Hide resolved
modules/gdscript/gdscript.cpp Outdated Show resolved Hide resolved
The function tried to rearrange properties but that lead to problems with duplication or deleted properties. Implemented the logic that that function did inside the get_property_list both for tool scripts and non-tool scripts.
@object71
Copy link
Contributor Author

object71 commented Aug 3, 2022

Thanks for the great suggestions and fixes. I added them in the latest version. Yeah I noticed the thing where it requires a reload of the scene to catch up some variables but it seems to be due to no call to update for the cached variables and I think it would be out of the scope of this PR. It is complicated enough as it is.

@akien-mga
Copy link
Member

I want this properly reviewed by @reduz but I expect this might take time to merge given the scope of the changes.

In the meantime, I was wondering if we should revert #58443 to fix the arguably "worse" regression in master / possibly alpha 14, until this new PR gets merged (and any changes from #58443 which are still needed can be added back here). WDYT?

@YuriSizov
Copy link
Contributor

OR! We could merge it for alpha 14 and let the public testing commence 😛

Up to you, but either way we have a rather prominent issue(s), whether we keep it as is or revert it.

@akien-mga
Copy link
Member

Well I wouldn't merge a core change without reduz even looking at it. Especially not as a follow-up to another core change which I merged without his review, which he then said was hacky, and which introduced a regression :)

@YuriSizov
Copy link
Contributor

To be fair, he said it was okay to merge it, just needed comments, and he called the method itself hacky, and the method was added in 2020, and I agree that it's a big hack.

@akien-mga
Copy link
Member

Well that's settled 😄

@akien-mga akien-mga merged commit 77d3ac7 into godotengine:master Aug 6, 2022
@akien-mga
Copy link
Member

Thanks!

@gb2dev
Copy link
Contributor

gb2dev commented Aug 6, 2022

Can someone check if this breaks #62936?

@YuriSizov
Copy link
Contributor

@bfeber I don't see why it would, the logic related to type_name which is used for the docs doesn't seem to be touched here.

@gb2dev
Copy link
Contributor

gb2dev commented Aug 6, 2022

It does break it, just tested and compared the GitHub Actions builds.

@YuriSizov
Copy link
Contributor

For future reference, the discussion continued in #63997.


#ifdef TOOLS_ENABLED
Ref<Script> script = get_script();
if (script->is_valid() && pcount > 0) {
Copy link
Contributor

@Gallilus Gallilus Jun 17, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Hi @object71 do you remember wy you use x->valid() in stead off x.valid()?
-> makes my GDExtension crash so before making a PR I want to check if it was intentional.

Copy link
Contributor

@Zylann Zylann Jun 17, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think it might need both (Ref::is_valid and Script::is_valid). The first checks if the object actually has a script attached to it. The second checks if the script itself is in a valid state. The code is currently missing the first.

Copy link
Member

@kleonc kleonc Jun 17, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It indeed should be script.is_valid() as it's the reference which needs to be validated before dereferencing it down below in script->get_class_category().


script.is_valid() will call:

inline bool is_valid() const { return reference != nullptr; }


script->is_valid() will dereference the stored reference pointer (possibly nullptr):

_FORCE_INLINE_ T *operator->() const {
return reference;
}

and call (possibly crashing):
virtual bool is_valid() const = 0;

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What do you mean by "possibly crashing" ?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What do you mean by "possibly crashing" ?

@Gallilus I mean if the Ref<Script> script's privateScript *reference member is indeed nullptr then script->is_valid() is equivalent to nullptr->is_valid(). And dereferencing nullptr is an undefined behavior, might result in a crash.

Hence before any script->... calls it should be ensured (e.g. by calling script.is_valid()) that it can be dereferenced.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This may sound silly but I'm not jet a regular in C++ conversations.
With dereference you mean going true the pointer in the object to do stuff.
Not deleting it from memory.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Also, I had a .ptr() in my code so ...

    MyClass my_object = memnew(MyClass);
    Ref<MyClass > reference_1 = *my_object;
    Ref<MyClass > reference_2a = reference_1;
    Ref<MyClass > reference_2b = reference_1.ptr();
    if (refrence_2b.is_valid()){
        // Is reference_2 a or b valid?
        // Will reference.is_valid() && refrence->is_valid() catch this problem?
    }

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
Archived in project
Development

Successfully merging this pull request may close these issues.

Exported variable appears doubled in object inspector