=proto
void duk_get_var(duk_context *ctx);

=stack
[ ... varname! ] -> [ ... val! ]

=summary
<p>Look up identifier <code>varname</code>.  Possible outcomes:</p>
<ul>
<li>If identifier is found, replaces <code>varname</code> with its value on the
    value stack and returns.</li>
<li>If identifier is not found, throws a <code>ReferenceError</code> (this applies
    to both strict and non-strict code).</li>
<li>If the input stack is empty (or some other internal error occurs),
    throws an error.</li>
</ul>

<p>The identifier lookup is equivalent to the Ecmascript expression:</p>
<pre class="ecmascript-code">
varname
</pre>

<p>For semantics, see
<a href="http://www.ecma-international.org/ecma-262/5.1/#sec-10.3.1">Identifier Resolution</a>,
<a href="http://www.ecma-international.org/ecma-262/5.1/#sec-10.2.2.1">GetIdentifierReference (lex, name, strict)</a>,
<a href="http://www.ecma-international.org/ecma-262/5.1/#sec-11.1.2">Identifier Reference</a>,
<a href="http://www.ecma-international.org/ecma-262/5.1/#sec-8.7.1">GetValue (V)</a>.
Ecmascript semantics require that a <code>ReferenceError</code> be thrown if the identifier
cannot be resolved (this is the case even in non-strict mode).
</p>

<div class="note">
Because Duktape/C functions cannot currently be nested functions, their
outer lexical environment is always the global environment.
</div>

=example
duk_push_string(ctx, "Math");
duk_get_var(ctx);  /* [ ... "Math" ] -> [ ... Math ] */
duk_get_prop_string(ctx, -1, "PI");  /* -> [ ... Math PI ] */
printf("Math.PI=%lf\n", duk_get_number(ctx, -1));
duk_pop_2(ctx);

=tags
variable
omit

=fixme
Although Ecmascript semantics always require a ReferenceError for
unresolvable identifiers, it might fit the API better to return zero
if identifier is not found and non-zero if found.  This would match
the property access API.  Incomplete.