Bug 396057

The following simple test structure definition for recognizing Pascal-style (run length ncoded) strings works only for strings longer than 1 byte, but for lengths 1 and 0 shows that the string has length 0 while still containing the previous content (sometimes with garbage characters at the beginning):

function init() {
	var obj = struct({
		len: uint16(),
		data: string("utf-8")
	});
	obj.byteOrder = "big-endian";
	obj.child("data").updateFunc = function() {
		this.maxByteCount = this.parent.len.value;
	};
	return obj;
}

Or the same expressed in the XML format:

<?xml version="1.0" encoding="UTF-8"?>
<data>
	<struct name="PascalString" byteOrder="big-endian">
		<primitive name="len" type="uint16" />
		<string name="data" encoding="utf-8" updateFunc="function() { this.maxByteCount = this.parent.len.value; }">
	</struct>
</data>

Here's a sample data I select:

00 04 41 42 43 44 45 46 47 48    |..ABCDEFGH    |

In this case, it correctly contains only a 4-byte string, "ABCD" (4).
But when I change the data to this:

00 01 41 42 43 44 45 46 47 48    |..ABCDEFGH    |

it contains a 0-byte string which is not empty! "BCDEFGH\0" (0).
It seems like it doesn't update the content of the string buffer / pointer to it, but updates the length to be 0 instead of 1.

A correct/expected behaviour should be a 1-byte string: "A" (1).
A 0-byte string should be a result when the length field in the data structure is set to `00 00`, and the contents of the string should be empty in that case, not contain the previously stored value: "" (0).

My version of Okteta is 0.12.5, which from some reason wasn't on the list. (KDE version 4.14.3)