can the UML be testable?

Recently i read someone promoting their own software modeling framework on some website. In explaining the advantages of his own system he remarked that UML "failed" because it was not testable or verifiable. The writer , however, is wrong in this assertion. I believe that the UML is testable and verifiable because (and assuming) it is directly mappable to source code, which is at least testable to the extent of the system requirements.

Now the idea of testability and verifiability is very important. and indeed these are two important criteria for well-designed software.

In a more general sense, when speaking of models in science, a crucial step in developing them is testing them, and their verifiability is also crucial in their acceptance and adoption.

So what about The UML? It is such a handy tool of design, it has become sort of the autoCAD or the HDL but for software instead of civil engineering or microprocessor design.

The UML to software is actually much more than what AutoCAD is to civic or mechanical engineering or HDL for chip design.

the UML is directly mappable to software, which can then be tested.

Whether the UML without code generation is itself testable might merit a web search.