The first problem is that the normal syntax check does not warn you of the incompatibility and the first you know is a short dump.
I am told (by Horst Keller) this is impossible, as the name of the function module is in quotes and so the normal syntax check cannot tell what function module it is dealing with, whereas a method call does not have the quotes so the syntax check knows what method it is dealing with and thus the signature. This also prevents you creating data using the new DATA construct when getting back parameters from a function module, whereas it works for a method.
I think that is a silly argument, as if the extended check can determine the exact function module then surely the normal syntax check can be enhanced to look in the same place, especially as we are talking about something which will 100% cause a short dump. My sneaky suspicion is that this is a way to try and stop us using function modules, despite the fact that sometimes there is no alternative (RFC, variant configuration etc.)
I have found that sometimes even passing in a variable with the same type and length as the expected parameter can cause a short dump, if the underlying data elements are different. Not always, but sometimes. No pattern that I can see.
In ABAP 7.40 and above you can wrap your parameter in the CONV construct and thus the CHAR value is converted into a string or vice versa and so no short dump. I use this very example in my good old book, as I recall, with a bit of a similar moan about the need for doing this in the first place.
Cheersy Cheers
Paul