Enum Ambiguity

Started by
3 comments, last by WitchLord 10 years, 4 months ago

I have noticed it is possible to register enum values with the same name as long as they belong to different enums, even though they have "global" scope. This, of course, can result in the same identifier resolving to multiple enum values. There is a compile error "Found multiple matching enum values" when the script writer attempts to use one of these enum values without a scope.

I believe in many of these circumstances, although there are multiple matching enum values, the enum value to use is not ambiguous. Consider the following example:


enum Enum1
{
   Yes,
   No
};

enum Enum2
{
    Yes,
    No,
    Maybe
};

void func()
{
    Enum1 e = Yes;
}

It should be possible to resolve the appropriate Enum1::Yes from the left-hand side of the assignment.

I say this knowing that you compile the right side before compiling the left side. I found this section "asCCompiler::CompileVariableAccess", as_compiler.cpp, line 7441:


// Look for the enum value without explicitly informing the enum type
asSNameSpace *ns = DetermineNameSpace(currScope);
int e = 0;
if( ns )
    e = builder->GetEnumValue(name.AddressOf(), dt, value, ns);
if( e )
{
    found = true;
    if( e == 2 )
    {
        Error(TXT_FOUND_MULTIPLE_ENUM_VALUES, errNode);
    }
}

The comment refers to what would be necessary to resolve the ambiguity. Is it possible to know the destination type at this point in the code?

At first glance, it looks like you would have to utilize a member in asSExprContext (similar to the use of "methodName") to hold the possible enum types for resolution later. This would probably be very much like resolving a object's implicit casts. Does this sound like a reasonable approach? Do you object in principle to the idea of resolving enum-value ambiguity, or might this be something worth doing? If I am on the right path, I think I could implement it myself.

Obviously, unique enum-value names or proper scoping resolves all ambiguity. My application's script writers are not experienced programmers though, so I do prefer to spare them from long enum-value names and scoping if I can help it.

Thank you for your consideration.

Advertisement

I'm definitely not against allowing the compiler to automatically determine which of the enums that was wanted when possible. To implement this support you will need to do something similar to what is done with functions and method names (i.e. the asSExprContext::methodName that you already found).

You're definitely on the right track, and if you can implement this I'll be glad to incorporate the changes in the compiler.

Regards,

Andreas

AngelCode.com - game development and more - Reference DB - game developer references
AngelScript - free scripting library - BMFont - free bitmap font generator - Tower - free puzzle game

I'm made the change, and it seems to work on my end. I ended up resolving the ambiguity in asCCompiler::ImplicitConvPrimitiveToPrimitive, but I didn't see a way to return from that function with a failure. Perhaps that wasn't a good place to do it, but it did involve a very small amount of code. I have attached my changes. Please let me know what you think.

I'll review the changes as soon as I can, and if I don't see anything wrong I'll have it included for the next release (2.28.0).

Thanks,

Andreas

AngelCode.com - game development and more - Reference DB - game developer references
AngelScript - free scripting library - BMFont - free bitmap font generator - Tower - free puzzle game

I've added this patch in revision 1784.

Thanks,

Andreas

AngelCode.com - game development and more - Reference DB - game developer references
AngelScript - free scripting library - BMFont - free bitmap font generator - Tower - free puzzle game

This topic is closed to new replies.

Advertisement