ObjC token annotations

[Jordan Rose]

I don't happen to know specifically, but I will note that "super" is not an expression; it's a context-sensitive keyword. ("self", on the other hand, is an implicit parameter with a fixed name.) As such, there's no useful Expr that can be attached to it. I would guess libclang picks the ObjCMessageExpr as the "best" context and leaves it at that.

Jordan


On Apr 2, 2013, at 12:44 , Erik Verbruggen <erikjv at me.com> wrote:

> Another token annotation question, this time about ObjC's self/super.
> 
> I have this snippet:
> 
>    SubIFace *inst;
>    [inst a];
>    [self a];
>    [super a];
> 
> When running that through c-index-test -test-annotate-tokens, I get some results I don't completely understand:
> 
> Identifier: "SubIFace" [153:14 - 153:22] ObjCClassRef=SubIFace:148:12
> Punctuation: "*" [153:23 - 153:24] VarDecl=inst:153:24 (Definition)
> Identifier: "inst" [153:24 - 153:28] VarDecl=inst:153:24 (Definition)
> Punctuation: ";" [153:28 - 153:29] DeclStmt=
> Punctuation: "[" [154:5 - 154:6] ObjCMessageExpr=a:146:10
> Identifier: "inst" [154:6 - 154:10] DeclRefExpr=inst:153:24
> Identifier: "a" [154:11 - 154:12] ObjCMessageExpr=a:146:10
> Punctuation: "]" [154:12 - 154:13] ObjCMessageExpr=a:146:10
> Punctuation: ";" [154:13 - 154:14] CompoundStmt=
> Punctuation: "[" [155:5 - 155:6] ObjCMessageExpr=a:146:10
> Identifier: "self" [155:6 - 155:10] DeclRefExpr=self:0:0
> Identifier: "a" [155:11 - 155:12] ObjCMessageExpr=a:146:10
> Punctuation: "]" [155:12 - 155:13] ObjCMessageExpr=a:146:10
> Punctuation: ";" [155:13 - 155:14] CompoundStmt=
> Punctuation: "[" [156:5 - 156:6] ObjCMessageExpr=a:146:10
> Identifier: "super" [156:6 - 156:11] ObjCMessageExpr=a:146:10
> Identifier: "a" [156:12 - 156:13] ObjCMessageExpr=a:146:10
> Punctuation: "]" [156:13 - 156:14] ObjCMessageExpr=a:146:10
> Punctuation: ";" [156:14 - 156:15] CompoundStmt=
> 
> So, "inst" in the first expression is a DeclRefExpr, just like "self" in the second one. However, "super" in the last expression is annotated with "ObjCMessageExpr=a"... Is that on purpose, and if so, what is the reasoning behind it?
> 
> Regards,
> Erik.
> _______________________________________________
> cfe-commits mailing list
> cfe-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits