Along this document there are several footnotes explaining or clarifying some decisions. Often these footnotes exposes the reason for changes in the language compared to the in-house compiler. These footnotes are purely informative and are not an integral part of this document.
Some changes demand more details than a footnote permits. These are marked with the footnote [1] and further details are presented in Appendix E.
+Some changes demand more details than a footnote permits. These are marked with the footnote [2] and further details are presented in Appendix E.
The lexical grammar of the language is context-sensitive. As such, the lexical elements and the syntactic elements are presented together.[2]
+The lexical grammar of the language is context-sensitive. As such, the lexical elements and the syntactic elements are presented together.[3]
Source code is a stream of printable ASCII characters plus the control codes line feed (\n), horizontal tab (\t) and carriage return (\r).[1]
Source code is a stream of printable ASCII characters plus the control codes line feed (\n), horizontal tab (\t) and carriage return (\r).[2]
Each line should be interpreted as if there is no whitespaces in either ends of the line.[4]
+Each line should be interpreted as if there is no whitespaces in either ends of the line.[5]
An integer literal is a sequence of digits optionally preceded by a minus sign.[1]
+An integer literal is a sequence of digits optionally preceded by a minus sign.[2]
If the literal begins with a minus, the number following it shall be negated.
@@ -712,7 +712,7 @@Once the F characters is found, all characters including and following it shall be ignored. The same shall happen when the character . is found a second time.[6]
Once the F characters is found, all characters including and following it shall be ignored. The same shall happen when the character . is found a second time.[7]
The literal can be preceded by a minus sign, which shall negate the floating-point number.
@@ -809,6 +809,9 @@String literals may not be supported by an implementation. In such case, a program using these literals is ill-formed.
+A variable reference is a variable name optionally followed by an array subscript.[1]
+A variable reference is a variable name optionally followed by an array subscript.[2]
The subscript uses an integer literal or another variable name of integer type for zero-based indexing.[7]
+The subscript uses an integer literal or another variable name of integer type for zero-based indexing.[8]
The program is ill-formed if the array subscript uses a negative or out of bounds value for indexing.
The program is ill-formed if a variable name is followed by a subscript but the variable is not an array.[1]
+The program is ill-formed if a variable name is followed by a subscript but the variable is not an array.[2]
An array variable name which is not followed by a subscript behaves as if its zero-indexed element is referenced.
@@ -848,6 +851,9 @@A variable used for indexing an array must not be an array.
If array variables are not supported, subscripts are ill-formed.
+A VAR_TEXT_LABEL parameter accepts an argument only if it is an identifier referencing a global variable of text label type.
+In case text label variables are not supported, this parameter type is not supported.
+A LVAR_TEXT_LABEL parameter accepts an argument only if it is an identifier referencing a local variable of text label type.
In case text label variables are not supported, this parameter type is not supported.
+A STRING parameter accepts an argument only if it is a string literal.
In case string literals are not supported, this parameter type is not supported.
+The INPUT_OPT parameter accepts an argument only if it is an integer literal, floating-point literal, or identifier matching a global string constant or referencing a variable of integer or floating-point type (in this order). A variable of text label type may be accepted by an INPUT_OPT parameter.
In case text label variables are not supported, the VAR_TEXT_LABEL_OPT and LVAR_TEXT_LABEL_OPT parameter types are not supported.
+An integer literal argument must have a parameter of type INT.
+An integer literal argument must have a parameter of type INT.
A floating-point literal argument must have a parameter of type FLOAT.
+A floating-point literal argument must have a parameter of type FLOAT.
For identifiers, the following applies (in the given order):
@@ -1081,16 +1099,16 @@If the identifier matches a global string constant, the parameter type must be INT and the argument shall behave as if rewritten as an integer literal corresponding to the string constant value.
If the identifier references a global variable, the parameter type must be either (depending on the type of the said variable) VAR_INT, VAR_FLOAT or VAR_TEXT_LABEL.
+If the identifier references a global variable, the parameter type must be either (depending on the type of the said variable) VAR_INT, VAR_FLOAT or VAR_TEXT_LABEL.
If the identifier references a local variable, the same rule as above applies, except by using LVAR_INT, LVAR_FLOAT and LVAR_TEXT_LABEL.
+If the identifier references a local variable, the same rule as above applies, except by using LVAR_INT, LVAR_FLOAT and LVAR_TEXT_LABEL.
If the identifier matches any string constant in any enumeration (except the global enumeration), the parameter type must be INPUT_INT and the argument shall behave as if rewritten as an integer literal corresponding to the string constant value.
+If the identifier matches any string constant in any enumeration (except the global enumeration), the parameter type must be INPUT_INT and the argument shall behave as if rewritten as an integer literal corresponding to the string constant value.
Otherwise, the parameter type must be TEXT_LABEL.
+Otherwise, the parameter type must be TEXT_LABEL.
SUB_THING_FROM_THING a c if the name a is the same as the name b.
Implementation-defined if a is the same name as c.[8]
Implementation-defined if a is the same name as c.[9]
SET a b followed by SUB_THING_BY_THING a c otherwise.
Constraints
The command it embodies cannot be any of the commands specified by this section (e.g. VAR_INT, ELSE, ENDWHILE, {, GOSUB_FILE, MISSION_START, etc).[1]
The command it embodies cannot be any of the commands specified by this section (e.g. VAR_INT, ELSE, ENDWHILE, {, GOSUB_FILE, MISSION_START, etc).[2]
An implementation may not support array variables. In such case, a program declaring a variable with dimensions is ill-formed.
+Constraints
The initial value of variables is unspecified.[9]
+The initial value of variables is unspecified.[10]
Constraints
The command it embodies cannot be any of the commands specified by this section (e.g. VAR_INT, ELSE, ENDWHILE, {, GOSUB_FILE, MISSION_START, etc).[1]
The command it embodies cannot be any of the commands specified by this section (e.g. VAR_INT, ELSE, ENDWHILE, {, GOSUB_FILE, MISSION_START, etc).[2]
The specified label location must be within a scope. Such scope may begin at the next non-empty embedded statement relative to the label location.[12]
+The specified label location must be within a scope. Such scope may begin at the next non-empty embedded statement relative to the label location.[13]
The type of a local variable and its respective input argument must match. For instance, if an input argument is an integer literal or variable of integer type, its corresponding local variable in the target scope must be of integer type.
@@ -3019,40 +3040,43 @@"This // is a string" is a string literal, not an incomplete string.
+5. This simplifies the syntactic specification.
"This // is a string" is a string literal, not an incomplete string.
NEGATE for substraction and ONEOVER for division) to perform this operation.
+8. The in-house compiler uses one-based indexing. Its array feature is incomplete and produces problematic bytecode when subscripting variables. Additionally, the GTA Vice City runtime (the target of the in-house compiler) does not support arrays. Thus it’s believed this feature is still incomplete in V413 and was never used. Not until GTA San Andreas, which introduces arrays in the execution environment. Its runtime performs zero-based indexing on variable subscripts, but for literal subscripts we can only guess. The compiled multifile contains debug strings that suggests zero-based indexing for literals.
NEGATE for substraction and ONEOVER for division) to perform this operation.
REPEAT in the in-house compiler disallows local variables. Due to being a complex command, however, the compiler ignores this parametric restriction and ends up permiting local variables (which is likely the intended behaviour).
+10. In Grand Theft Auto, the memory storage of local variables is not zeroed after leaving a scope (nor while entering one). This means local variables preserve values from previous scopes. Additionally, the Stories chapters share the storage of global variables between mission scripts, having the very same implications (no in-depth research has been conduced though).
SCRIPT_NAME, but this does not tell much because it barely supports any text label variable feature. No compiled multi-file contains script names assigned by a variable. Thus, we refrain from defining any semantics for this case.
+11. The parameter definition of REPEAT in the in-house compiler disallows local variables. Due to being a complex command, however, the compiler ignores this parametric restriction and ends up permiting local variables (which is likely the intended behaviour).
SCRIPT_NAME, but this does not tell much because it barely supports any text label variable feature. No compiled multi-file contains script names assigned by a variable. Thus, we refrain from defining any semantics for this case.
+