jaydenmilne / tib Goto Github PK

Right now a new layer of recursion is added for each line that is parsed. This is wasteful, this functionality could be accomplished with a loop. This will mean paring down the grammar a bit.

Expected:
{1, 2} or {4, 0} gives {1 0}
Actual:
{1, 2} or {4, 0} gives 1

Disp should be rudimentially implemented, instead of always outputting the result of executing a line.

When a program requests a variable that has not been set, VariableManager::get initializes a Value without providing a value (Value defaults to an uninitialized double) and returns it to make initialization errors obvious. However, get should at least check the type of the variable and return garbage data of the correct type, so Str1 at least returns a garbage string and not a double.

The Value class is terrible. If we used some inheritance scheme, we might have to do runtime type checking, but we would be able to be much more dynamic

When parsing a string tib will continue until it finds an apostrophe or the end of the line. TI-BASIC will parse a string until it finds a token with higher precedence,

In essence, tib will parse
"burrito->Str1 as the string "burrito->Str1" while TI-BASIC will interpret it as "burrito"→Str1 and will set Str1 to "burrito".

Since tib uses ->, a hyphen and a greater than symbol, to represent the STO character → it would be difficult to recognize the combination -> as a string terminator, since then a string could not contain -> without escaping it somehow, which would be annoying.

Research needs to be done to see what other symbols have this different precedence. Maybe an alternate -> symbol could be used?

Expected:
-{1,2} gives {-1, -2}
Actual:
-{1,2} gives ERR:DATA TYPE

VariableManager::set when in emulation mode (-e) should detect the type of the variable and make sure that the value being assigned to it is of the correct type, eg "asdf"->X and 1->Str1 gives an error

ANS needs to be implemented