We translate our website in various languages. Each page of the website contains many URL-s to other pages, but the URL-s are not known by the translator.
In fact, they first have to translate the names of the pages, these get loaded into the CMS, which establishes the new URL-s (which may change later), and then someone manually adds them to the translation memory using 'vi' or notepad++.
However, we also have generated a TMX that contains the then-valid translations. But since the URL-s have already been translated, these are not picked automatically.
It would save many hours of work and testing if we had some kind of TMX folder that contains TMX-es that would be applied automatically PLUS when a translation already exists would REPLACE the existing translation.
It could also be used for other purposes, such as enforcing that some sentences are never translated. For instance, a segment name with a company name that might be interpreted as a normal English word (such as "King"). Of course, a glossary helps but this enforces it.