This a bug in toke.c, in the following code:

STATIC void
S_checkcomma(pTHX_ register char *s, char *name, char *what)
{
    char *w;

    if (*s == ' ' && s[1] == '(') {     /* XXX gotta be a better way *
+/
        if (ckWARN(WARN_SYNTAX)) {
            int level = 1;
            for (w = s+2; *w && level; w++) {
                if (*w == '(')
                    ++level;
                else if (*w == ')')
                    --level;
            }
            if (*w)
                for (; *w && isSPACE(*w); w++) ;
            if (!*w || !strchr(";|})]oaiuw!=", *w))     /* an advisory
+ hack only... */
                Perl_warner(aTHX_ WARN_SYNTAX,
                            "%s (...) interpreted as function",name);
        }
    }
    # [the S_checkcomma function continues from here with an unrelated
+ syntax check]
[download]

 

Here's the code again, this time with comments interspersed:

STATIC void
S_checkcomma(pTHX_ register char *s, char *name, char *what)
{
    char *w;

    if (*s == ' ' && s[1] == '(') {     /* XXX gotta be a better way *
+/
[download]

When this function is called, the tokenizer has just grabbed the name of a function. If the current character is a space, and the next character is an open paren, then we'll do the following...         if (ckWARN(WARN_SYNTAX)) { (...but only if syntax warnings are turned on...)

            int level = 1;
            for (w = s+2; *w && level; w++) {
                if (*w == '(')
                    ++level;
                else if (*w == ')')
                    --level;
            }
[download]

That for loop should be familiar to many programmers. It's simply counting open and close parens. The conditional, *w && level, will cause the loop to exit either at the end of the line (when *w is null) or when the matching close paren for the first open paren is found (level is 0). However, note that the loop ignores anything that isn't a paren.

            if (*w)
                for (; *w && isSPACE(*w); w++) ;
[download]

This extra loop just skips any whitespace after that last closing paren.             if (!*w || !strchr(";|})]oaiuw!=", *w))     /* an advisory hack only... */ Now, if we're at the end of the line (*w is null) or the current character is not one of ;|})]oaiuw!=, then the warning in question is produced:

                Perl_warner(aTHX_ WARN_SYNTAX,
                            "%s (...) interpreted as function",name);
        }
    }
[download]

Remember the part about the counting of parens ignoring anything that isn't a paren? So, in print ("hey)");, this bit of toke.c thinks that the first close paren is the matching paren, even though that paren is inside a string. The character after that paren is a double quote, which is not in that set of acceptable characters, and so the warning is produced.

This suggests some other interesting snippets:

print (/\)/);      # warning
print (/./);       # no warning
print ("hey))");   # no warning
print ("hey")&&1;  # warning
print ("hey")||1;  # no warning
[download]

As the comment says, there's "gotta be a better way". :)

To avoid this warning, you could remove the space before the open paren, add a plus before the open paren, or remove the parens entirely if they're unneeded.

By the way, you're probably wondering why the set of characters ;|})]oaiuw!= is special. Frankly, I just don't know!