Translator – Temporary String Support - Debugging

The Parser tests expected results needed to be updated because the EOL code had changed (due to extra table entries being added). Several of the Translator tests expected results needed to be updated for the new temporary string codes. For example, the convention for the operators are CatStr: +$, CatStrT1: +$1, CatStrT2: +$2, and CatStrTT: +$T. The convention for functions is a T added to the function name, for example Len: LEN( and LenTmp: LENT(. The INSTR functions with two string arguments use the convention T1, T2 and TT.

Some minor table entries problems were fixed, but one significant problem that needed correcting was in the number of strings member in the expression information structure in each table entry. This member is automatically calculated during table initialization, but was counting both string and temporary string operands. This field is used to determine how many strings need to be popped from the done stack and attached. Temporary strings are not left on the done stack, so the number of strings should have only included the number of strings only, not temporary strings.

There was one other issue with constant strings. Technically these are already known to be strings, unlike tokens with and without parentheses that may be variables/arrays or user functions. Therefore, constants don't need to be attached to operands. But there is no allowance in the current implementation for this, so to keep things simple, they will continue to be attached. The proper code is already set (for a string operand), and the Encoder will see that they are constants and leave the code as is. Normally, if the Encoder determines that an operand is user function, meaning it produces a temporary string, it will change the code appropriately to one that has a temporary string operand.

There is still a problem with assignments...