Also in QFG2... The decompiler produces a bunch of "--invalid--" tokens in Said strings. It turns out (of course), that these *are* valid. They just correspond to duplicate words in the vocab resource.
For example, QFG2 has:
word group 451: open verb
word group 1246: open verb/noun
When Companion loads the vocab resource, it assembles a dictionary of word -> group, and assumes each word is unique. So one word goes missing, and the decompiler can't map from word group to word.
So the questions are:
- Is that intended behavior for Sierra to have duplicate words? Was it used for different words in other languages maybe? (someone could probably test this)
- How did their compiler disambiguate them? 'open, open>' -> how does it know the first open is 451, and the second one is 1246?
