[m-users.] Determinism error caused by instantiation error, not sure why.

[Zoltan Somogyi]

2022-10-31 10:11 GMT+11:00 "Sean Charles (emacstheviking)" <objitsu at gmail.com>:
> The error:
> 
> repl.m:738: In `take_gencall'(in, out, in, out, in, out):
> repl.m:738:   error: determinism declaration not satisfied.
> repl.m:738:   Declared `det', inferred `semidet'.
> repl.m:747: In clause for `take_gencall(in, out, in, out, in, out)':
> repl.m:747:   warning: unification of `Out' and maybe.yes cannot succeed.
> repl.m:747:   `Out' has instantiatedness `unique(no)'.
> 
> The code:
> 
> :- pred take_gencall(error_msg::in, maybe(snode)::out, s::in, s::out,
>     lsnode::in, lsnode::out) is det.
> 
> take_gencall(Er, Out, !S) -->
>     ( if no_errors(!.S) then
>         ( if [GC] then
>             { Pos = snpos(GC) },
>             ( if { GC = gencall(_, _, _) } then
>                  { Out = yes(GC) }   <== line 747
>             else
>                 cerror(Pos, Er, !S),
>                 { Out = no }
>             )
>         else
>             premature(!S)),
>             { Out = no }
>     else
>         { Out = no }
>     ).
> 
> For the life of me I don't understand why the line in bold is causing an error, why has Out become `unique(no)` and does that no refer to a general no or is it the no from the may() discriminated union, I just don't see my mistake. Sorry.

Because the code you are looking at is not what the compiler sees.

You have a misplaced close parenthesis. It *should* be just before the final "else",
but in actuality, it occurs just after "premature(!S)". So what the compiler sees
is a conjunction containing both (a) the code after the call to premature,
which is Out = no, and the if-then-else containing Out = yes(GC).

Zoltan.