A method to achieve executability of temporal logic specifications is restricting the interpretation of formulas over finite domains. This method was proven to be successful in designing testing tools used in industrial applications. However, the extension of the results of these tools to infinite, or just larger, domains, requires an adequate definition of a “finite-domain semantics”. Here we show the need for correcting previous semantics proposals, especially in the case of specifications of real-time systems which require the use of bounded temporal operators. We define a new semantics for the TRIO language, easily extendible to other linear metric temporal logic languages and show its adequateness on various examples.