Aleksandar Donev wrote: > Intel's compiler gives an error for the following function because it > does not assign a return result: > > function test(x) > write(*,*) "error!" > stop > end function test > > Is this kind of compiler behavior allowed? If yes, why should a silly > compiler be allowed to play so smart (it should of course give a > warning)... actually, the silly compiler is completely correct - the function result *must* be assigned (or otherwise given a value), even if the silly function is just going to print a msg and stop -John Turner LANL, CCS-2