Stringnames without double quotes

[Agecaf]

If I understand correctly, StringNames have a literal form in GDScript, so writing &"ui_accept" generates the StringName for accepting ui. What I'm thinking is that it might be nice to be able to also write &ui_accept to generate the same StringName.

StringNames are often used for strings that do not have whitespace inside them, like names of methods in Object.call or action names in Input.is_action_just_pressed.

Currently writing &ui_accept syntax highlights it the same color as a proper StringName like &"ui_accept", even if &ui_accept is a syntax error. So from a parsing perspective it should be a relatively simple addition. This also means that if you forget you need double quotes on StringNames, you run into a syntax error

There may be confusion with the binary & operator, but usually string names, like identifiers, do not start with a digit, and the syntax highlighter seems fairly good at detecting when an & is being used as a binary operator or as the start of a StringName. If string names starting with digits or with spaces are required, users would have no issue using the previous notation as &"42" or &"Hello World!".

This behavior would be similar to the not-literal $ for accessing nodes, which can be currently used as $ChildNode or $"ChildNode" interchangeably.

[dalexeev]

See:

• Improve StringName and NodePath usability #4721
• Simplified StringName literal for single word text #8079

[Mickeon]

Actually, see dalexeev's own comment on the matter in #4721 (comment)

[Agecaf]

Thanks for the clarification! I wasn't able to find these discussions in my first search. I thought the reason that the StringName operator was chosen to be like &"this" (as opposed to something like s"this" similar to r"raw strings") was to make it similar to pointers/references in C, since StringNames are often used as references to method names and input action names.

StringNames if I understand correctly are the equivalent of keywords in Clojure or symbols in Ruby, both of which use a notation like :this (with Ruby also having a notation like :"this"), they don't seem to require the double quotes in those cases from a usability perspective.

I think there's a usability for users familiar with StringNames vs confusion for users not-yet-familiar with StringNames balance to be found, but this could be addressed from a documentation perspective. Nodepaths like $this made me initially confused that GDScript was going to be like Perl, but that was quickly dispelled once I learnt what Nodepaths were and how the notation was super useful.

[dalexeev]

Another important difference between the $Node construct and the &"StringName" and ^"NodePath" literals is that if you make a typo in the node name, you'll get a runtime error immediately, whereas strings can contain any content and the error may occur much later.

If you see &my_method, you might incorrectly think that this syntax provides some compile-time or at least runtime guarantees that my_method actually exists (like first-class callables and signals do). See also nameof proposals (#837, #10117). In my opinion, the quotes more explicitly tell you "hey, I'm just a string (albeit a special one) and can contain anything".