Tag: 2014

Originally posted on the Unreal Engine blog.

A Blueprint Macro lets you reuse a set of nodes over and over, and can be created in any blueprint (using the Add Macro button on the ‘My Blueprint’ toolbar). You can also turn a selection into a macro by right-clicking on a selected node and using the ‘Collapse to Macro’ option. A macro works a lot like a collapsed graph; you can define arbitrary inputs and outputs, which will show up as pins whenever you place a macro instance.  Here is an example of a macro named IsValid, which checks to see if an object pin is valid or not:

This macro can then be placed as an instance in another graph as ‘shorthand’ for the nodes it contains, allowing you to reuse code and hide complexity:

Continue reading “Blueprint Macros and Macro Libraries”

Originally posted on the Unreal Engine blog.

As you build larger projects with Blueprints, it’s easy to end up with an overwhelming sea of nodes.  However, we’ve built in a number of different encapsulation and code reuse mechanisms to help you battle the chaos.

To encapsulate something is “to show or express the main idea or quality of (something) in a brief way”; in other words we can hide a complex sequence of nodes with a simple stand-in that conveys the same meaning or idea.  You can still drill down and see how it works ‘under the hood’, but you don’t have to worry about the details when looking at the broader picture. Continue reading “Managing complexity in Blueprints”

Originally posted on the Unreal Engine blog.

We’ve worked really hard to provide you guys with lots of helpful shortcuts to streamline your workflow within the Editor. One of those improvements is the addition of asset navigation shortcuts.

Opening an asset to edit

You can Pick an asset to edit from almost anywhere in the editor by using the Ctrl+P shortcut. This will open an asset picker with the search box already focused. Once you’ve filtered the results, you can use the arrow keys to select an asset and hit Enter to start editing it. The mouse also works to select or scroll, and you can even drag/drop out of the picker into the level editor.

Working with the assets of selected actors

When you have one or more selected actors in the level editor, you can Edit their associated assets using Ctrl+E. For example, a selected static mesh actor will open the associated static mesh, while a selected blueprint instance will open the blueprint.

If you want to perform some other action on the asset, or just find out where it lives in the Content tree, you can use Ctrl+B to select the associated asset in the Content Browser. This shortcut also works in any asset editor, selecting the current asset being edited in the content browser.

Originally posted on the Unreal Engine blog.

So, source control is awesome, and you should be using it! One of the benefits of source control is being able to look at previous versions of files and compare or ‘diff’ them to see how things have changed. However, regular diff tools don’t work very well on binary files like Unreal .uasset packages. We’ve got you covered with support for exporting a text version of assets that can be diffed using your preferred tool, as well as a built-in specialized diff tool for Blueprints.

Setup

You’ll need to store your project in a supported source control system for the built-in tools to work. The editor currently has source control plugins for both Perforce and SVN. Perforce is what we use internally at Epic, and it is now free for up to 20 users (and 20 workspaces; you’ll probably end up using 2 or 3 per human user). SVN is totally free and there are a number of cloud SVN providers if you don’t want to host your own server.

Note: SVN binaries were not included in the 4.0 release, check out this thread for more details.

You can tell if your editor is currently connected to source control by the little icon on the top right of the main frame, next to the [Enter console command] prompt. If the icon is green, you are already connected, but if it shows up as a red no sign then click on it to enter your server settings.

Next, you need to tell the editor where it can find your favorite text diff tool in Editor Preferences. It defaults to p4merge, but almost any diff tool should work (e.g., Beyond Compare, Araxis Merge, Tortoise Merge, etc…)

Diffing assets

Once you are connected, you’ll see additional options in the context menu for Assets, such as ‘Check Out’, ‘History’, and ‘Diff Against Depot’. Any asset can be diffed, and the default handler will export the asset to a readable text format and launch your preferred text diffing tool (some assets have a more sophisticated viewer). Comparing your current version against depot before checking in is a useful habit to form; it will help you write better checkin comments and can also help catch changes you made to test something but didn’t intend to keep.

History gives you a full list of the previous checkins and their descriptions, and you can diff any two arbitrary revisions from the history window (select two revisions and right-click). This can be extremely useful when trying to track down a regression (be kind to future you and write good checkin comments).

Diffing Blueprints

Blueprints are a much more visual medium and a textual diff wouldn’t be very helpful. The editor has a built-in diffing tool for blueprints where you can look at differences in each graph or diff the default values. We’ll talk about this in more detail in a future blog post, but here’s a picture of the tool

Originally posted on the Unreal Engine blog.

Reflection is the ability of a program to examine itself at runtime. This is hugely useful and is a foundational technology of the Unreal engine, powering many systems such as detail panels in the editor, serialization, garbage collection, network replication, and Blueprint/C++ communication. However, C++ doesn’t natively support any form of reflection, so Unreal has its own system to harvest, query, and manipulate information about C++ classes, structs, functions, member variables, and enumerations. We typically refer to reflection as the property system since reflection is also a graphics term.

The reflection system is opt-in. You need to annotate any types or properties that you want to be visible to the reflection system, and Unreal Header Tool (UHT) will harvest that information when you compile your project.

Markup

To mark a header as containing reflected types, add a special include at the top of the file. This lets UHT know that they should consider this file, and it’s also required for the implementation of the system (see the ‘A peek behind the curtain’ section for more information).

#include "FileName.generated.h"

You can now use UENUM(), UCLASS(), USTRUCT(), UFUNCTION(), and UPROPERTY() to annotate different types and member variables in the header. Each of these macros goes before the type or member declaration, and can contain additional specifier keywords. Let’s take a look at a real world example (from StrategyGame):

// Base class for mobile units (soldiers)
 
#include "StrategyTypes.h"
#include "StrategyChar.generated.h"
 
UCLASS(Abstract)
class AStrategyChar : public ACharacter, public IStrategyTeamInterface
{
	GENERATED_UCLASS_BODY()
 
	/** How many resources this pawn is worth when it dies. */
	UPROPERTY(EditAnywhere, Category=Pawn)
	int32 ResourcesToGather;
 
	/** set attachment for weapon slot */
	UFUNCTION(BlueprintCallable, Category=Attachment)
	void SetWeaponAttachment(class UStrategyAttachment* Weapon);
 
	UFUNCTION(BlueprintCallable, Category=Attachment)
	bool IsWeaponAttached();
 
protected:
	/** melee anim */
	UPROPERTY(EditDefaultsOnly, Category=Pawn)
	UAnimMontage* MeleeAnim;
 
	/** Armor attachment slot */
	UPROPERTY()
	UStrategyAttachment* ArmorSlot;
 
	/** team number */
	uint8 MyTeamNum;
 
	[more code omitted]
};

This header declares a new class called AStrategyChar deriving from ACharacter. It uses UCLASS() to indicate it is reflected, which is also paired with a macro GENERATED_UCLASS_BODY() inside of the C++ definition. The GENERATED_UCLASS_BODY() / GENERATED_USTRUCT_BODY() macros are required in reflected classes or structs, as they inject additional functions and typedefs into the class body.

The first property shown is ResourcesToGather, which is annotated with EditAnywhere and Category=Pawn. This means the property can be edited in any details panel in the editor, and will show up in the Pawn category. There are a couple of annotated functions marked with BlueprintCallable and a category, meaning they’ll be available to call from Blueprints.

As the MyTeamNum declaration shows, it’s fine to mix reflected and non-reflected properties in the same class, just be aware that the non-reflected properties are invisible to all of the systems that rely on reflection (e.g., storing a raw unreflected UObject pointer is usually dangerous since the garbage collector can’t see your reference).

Each of the specifier keywords (such as EditAnywhere or BlueprintCallable) is mirrored in ObjectBase.h with a short comment on the meaning or usage. If you’re not sure what a keyword does, Alt+G will usually work to take you to the definition in ObjectBase.h (they aren’t real C++ keywords, but Intellisense or VAX don’t seem to mind/understand the difference).

Check out the Gameplay Programming Reference for more information.

Limitations

UHT isn’t a real C++ parser. It understands a decent subset of the language and actively tries to skip any text that it can; only paying attention to reflected types, functions, and properties. However, some things can still confuse it, so you may have to reword something or wrap it in an #if CPP / #endif pair when adding a reflected type to an existing header. You should also avoid using #if/#ifdef (except for WITH_EDITOR and WITH_EDITORONLY_DATA) around any annotated properties or functions, since the generated code references them and will cause compile errors in any configuration where the define isn’t true.

Most common types work as expected, but the property system can’t represent all possible C++ types (notably only a few template types such as TArray and TSubclassOf are supported, and their template parameters cannot be nested types). UHT will give you a descriptive error message if you annotate a type that can’t be represented at runtime.

Using reflection data

Most game code can ignore the property system at runtime, enjoying the benefits of the systems that it powers, but you might find it useful when writing tool code or building gameplay systems.

The type hierarchy for the property system looks like this:

• UField
  • UStruct
    • UClass (C++ class)
    • UScriptStruct (C++ struct)
    • UFunction (C++ function)
  • UEnum (C++ enumeration)
  • UProperty (C++ member variable or function parameter)
    • (Many subclasses for different types)

UStruct is the basic type of aggregate structures (anything that contains other members, such as a C++ class, struct, or function), and shouldn’t be confused with a C++ struct (that’s UScriptStruct). UClass can contain functions or properties as their children, while UFunction and UScriptStruct are limited to just properties.

You can get the UClass or UScriptStruct for a reflected C++ type by writing UTypeName::StaticClass() or FTypeName::StaticStruct(), and you can get the type for a UObject instance using Instance->GetClass() (it’s not possible to get the type of a struct instance since there is no common base class or required storage for structs).

To iterate over all members of a UStruct, use a TFieldIterator:

for (TFieldIterator<UProperty> PropIt(GetClass(), EFieldIteratorFlags::IncludeSuper); PropIt; ++PropIt)
{
	UProperty* Property = *PropIt;
	// Do something with the property
}

The template argument to TFieldIterator is used as a filter (so you can look at both properties and functions using UField, or just one or the other). The second argument to the iterator constructor indicates whether you only want fields introduced in the specified class/struct, or fields in the parent class/struct as well (the default); it doesn’t have any effect for functions.

Each type has a unique set of flags (EClassFlags + HasAnyClassFlags, etc…), as well as a generic metadata storage system inherited from UField. The keyword specifiers are usually either stored as flags or metadata, depending on whether they are needed in a runtime game, or only for editor functionality. This allows the editor-only metadata to be stripped out to save memory, while the runtime flags are always available.

You can do a lot of different things using the reflection data (enumerating properties, getting or setting values in a data-driven manner, invoking reflected functions, or even constructing new objects); rather than go in depth on any one case here, it’s probably easier to have a look thru UnrealType.h and Class.h, and track down an example of code that does something similar to what you want to accomplish.

A peek behind the curtain

You can safely skip this section if you just want to use the property system, but knowing how it works helps motivate some of the decisions and limitations in headers that contain reflected types.

Unreal Build Tool (UBT) and Unreal Header Tool (UHT) act in concert to generate the data that is needed to power runtime reflection. UBT has to scan headers to do its job, and it remembers any modules that contain a header with at least one reflected type. If any of those headers have changed since the last compile, UHT is invoked to harvest and update the reflection data. UHT parses the headers, builds up a set of reflection data, and then generates C++ code containing the reflection data (contributing to a per-module .generated.inl), as well as various helpers and thunk functions (per-header .generated.h).

One of the major benefits of storing the reflection data as generated C++ code is that it is guaranteed to be in sync with the binary. You can never load stale or out of date reflection data since it’s compiled in with the rest of the engine code, and it computes member offsets/etc… at startup using C++ expressions, rather than trying to reverse engineer the packing behavior of a particular platform/compiler/optimization combo. UHT is also built as a standalone program that doesn’t consume any generated headers, so it avoids the chicken-and-egg issues that were a common complaint with the script compiler in UE3.

The generated functions include things like StaticClass() / StaticStruct(), which make it easy to get reflection data for the type, as well as thunks used to call C++ functions from Blueprints or network replication. These must be declared as part of the class or struct, which explains why a GENERATED_UCLASS_BODY() or GENERATED_USTRUCT_BODY() macro is included in your reflected types, as well as the #include "TypeName.generated.h" which defines those macros.