Well, your example is a bit far-fetched. Normally, the hidden non-translatable fragments give some context in the segment which might be important to neural engines in MT. Apart from it, most of the time the MT translation includes the non-translatable fragments so that you don't need to transfer them back manually.
CafeTran Espresso doesn't provide a filter for Schema4 XML. So I have to run them via Studio. This means that nearly every segment starts with a bullet (ü or 4) and a tag.
So, for me this isn't that far-fetched.
I have defined this hidden non-translatable:
And it is left out of the segment in the Segment Editors beautifully.
However, it looks like the segment in its long form, so with the hidden non-translatable, is sent to the MT systems. Clearly, this is not what I want ...
Can this be fixed, please?