Incorrect error message when same identifier used for both fct name and var name

[ksvanhorn]

Summary:

When the same identifier is used for both a fct name and var name, it misidentifies this as use of a reserved word.

Description:

I use "foo" as the name of a function in the functions block. Then I also use it as a parameter name in the parameters block. The error message says "variable identifier (name) may not be a reserved word", but "foo" is not a reserved word.

Reproducible Steps:

code <- '
functions {
  real foo(real x) {
    return 2.0 * x;
  }
}
data { int<lower=0> N; vector[N] y; }
parameters {
  real foo;
}
model {
  y ~ normal(foo, 1);
}
'
library(rstan)
sm <- stan_model(model_code = code)

Current Output:

SYNTAX ERROR, MESSAGE(S) FROM PARSER:

variable identifier (name) may not be reserved word
    found identifier=foo
  error in 'model1219268b0a499_bf344b66bb761e274281c4862a3bdfdd' at line 9, column 7
  -------------------------------------------------
     7: data { int<lower=0> N; vector[N] y; }
     8: parameters {
     9:   real foo;
              ^
    10: }
  -------------------------------------------------

PARSER EXPECTED: <identifier>
Error in stanc(file = file, model_code = model_code, model_name = model_name,  : 
  failed to parse Stan model 'bf344b66bb761e274281c4862a3bdfdd' due to the above error.

Expected Output:

I expect it to complain that "foo" is already being used as the name of a function.

Additional Information:

Current Version:

v2.17.3

[bob-carpenter]

Definitely a bug in the error message. We should be able to reuse the same term for a function name and variable, too. We'll look into it and fix. Thanks for reporting.

[VMatthijs]

Stanc3 now says

Semantic error at file "test.stan", line 8, characters 7-10:
   -------------------------------------------------
     6:  data { int<lower=0> N; vector[N] y; }
     7:  parameters {
     8:    real foo;
                ^
     9:  }
    10:  model {
   -------------------------------------------------

Identifier  foo  is already in use.

[bob-carpenter]

1. There's too much space around "foo" in this example. 2. Should we be quoting identifiers in error messages? 3. Should we say "column" rather than "characters"? For instance, in clang++, this program: int main() { double foo; double foo; return 1; } produces this error: cpperr.cpp:3:10: error: redefinition of 'foo' double foo; ^ cpperr.cpp:2:10: note: previous definition is here double foo; ^ 1 error generated. I'm not saying we need to go back to the original definition as this does! The line numbers and columns aren't even marked, just 2:10 for line 2, column 10. I like that you're spelling it out. - Bob

…

On Dec 13, 2018, at 7:34 AM, Matthijs Vákár ***@***.***> wrote: Stanc3 now says Semantic error at file "test.stan", line 8, characters 7-10: ------------------------------------------------- 6: data { int<lower=0> N; vector[N] y; } 7: parameters { 8: real foo; ^ 9: } 10: model { ------------------------------------------------- Identifier foo is already in use. — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.

[VMatthijs]

Thanks for the feedback! I agree. I've fixed points 1-3 in stanc3.

[mcol]

Current message is:

Semantic error in 'test.stan', line 8, column 7 to column 10:
   -------------------------------------------------
     6:  data { int<lower=0> N; vector[N] y; }
     7:  parameters {
     8:    real foo;
                ^
     9:  }
    10:  model {
   -------------------------------------------------

Identifier 'foo' is already in use.