parse(#Exp, "2+3") example doesn't work

[Pytheas01]

Describe the bug

the parse(#Exp, "2+3") statement causes a CallFailed error.

To Reproduce

rascal>import demo::lang::Exp::Concrete::NoLayout::Syntax;
ok
rascal>import ParseTree;
ok
rascal>parse(#Exp, "2+3");

causes

rascal>parse(#Exp, "2+3");
|prompt:///|(12,5,<1,12>,<1,17>): CallFailed(
  |prompt:///|(12,5,<1,12>,<1,17>),
  [type(
      adt(
        "Exp",
        []),
      (adt(
          "Exp",
          []):choice(
          adt(
            "Exp",
            []),
          {
            cons(
              label(
                "add",
                adt(
                  "Exp",
                  [])),...

Expected behavior

sort("Exp"): 2+3

... as per the documentation

Desktop (please complete the following information):
Eclispse 2020.03 using 2020-May-13 version of Rascal update.

[Pytheas01]

This morning, I've managed to make the example work, as follows:

• right-click on project and select "Rascal Console"
• import Concrete::NoLayout::Eval
• import Concrete::NoLayout::Syntax
• import ParseTree
  Only then can you type eval(parse(#Exp, "2+2")) at the prompt.
  I wonder why, after importing the Eval module, you have to import the Syntax and ParseTree modules again?

[jurgenvinju]

Thanks for the report!

The issue is caused by the two different versions of Exp both in scope in the console. The current interpreter does not report the ambiguity and picks the first it can find for #Exp. In your case that was the data version and so there was no definition for parse to match.

The upcoming statically checked version of Rascal will warn about this and force a more qualified name.

Your workaround works accidentally; caused by the same bug: now you have the syntax version of Exp being resolved for #Exp and it goes well.

The best way to work with this for now is to separate scopes that import the concrete syntax from the scopes that use the abstract syntax. And thanks for the patience while we work on the Compiler-based REPL!

[jurgenvinju]

Thanks for the diagnosis in Slack @tvdstorm

[jurgenvinju]

Does this analysis help you @PHIROC ?

Note to self: change the instructions in the documentation to avoid this pitfall.

[Pytheas01]

Hi Jurgen, Yes it did. Thank you. Best regards, Phiroc

…

Le 24 mai 2020 à 08:32, Jurgen J. Vinju ***@***.***> a écrit :  Does this analysis help you @PHIROC ? Note to self: change the instructions in the documentation to avoid this pitfall. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.

[jurgenvinju]

This will be fixed by #1835