Global variables #378
Global variables #378
Conversation
There was a problem hiding this comment.
A light pass of comments. This generally seems reasonable to me, but it's probably worth not considering these globals... Although it hasn't been implemented in executable_semantics, everything in Carbon should be in a package (https://github.com/carbon-language/carbon-lang/tree/trunk/docs/design/code_and_name_organization), and so these are package-scoped variables, not global.
| @@ -130,6 +131,20 @@ struct ChoiceDeclaration { | |||
| auto TopLevel(ExecutionEnvironment&) const -> void; | |||
| }; | |||
|
|
|||
| // Global variable definition implements the Declaration concept. | |||
There was a problem hiding this comment.
I'm a bit confused trying to read this comment, can you try rephrasing a little? Maybe something like "Package variable definitions use the declaration concept, even though they aren't strictly declarations", if that's accurate?
There was a problem hiding this comment.
I thought that definitions can also be declarations. But perhaps Carbon is going with different terminology? Or perhaps I've miss remembered the terminology. In any event, the resolution to this would perhaps yield a name change to the Declaration class.
There was a problem hiding this comment.
To be clear:
- I think this comment isn't using correct grammar (e.g., no "the" at the start, which implies "Global variable definition" is a specific noun).
- "Global variable definition" refers to a concept that shouldn't exist in Carbon.
- I'm also not clear what the "Declaration concept" refers to when "Declaration" is capitalized (if it's a subclass, why isn't it one? I may just be a bit confused by the design of this file)
|
@jonmeow Regarding global vs. package scope... yes, these variables will be package scope once packages are added to the executable semantics. |
* global variables * implemented type checking of global variable, added test case * added a comment * improvements based on Dave's suggestions * improvements based on Jon's suggestions * added test cases about global variable ordering
This PR implements global variables with the syntax
var type : name = expression ;
The motivation to put this in now is that it introduces a feature with side effects that are visible outside of a given function call. Having this will allow us to test the order of evaluation of other language features, such as the experimental delimited continuations.