Features

Feature Comparison

Non-Features

• Many concepts of object-oriented programming languages are not supported, for example inheritance, method overloading, polymorphism, and more complex encapsulation.
• Many concepts of functional programming languages are not supported, for example high-order functions, functional composition, closures. This is on purpose, to avoid choice friction, and to simplify learning and using the language.
• Reflection is not supported.
• Tail calls are only optimized by the C compiler.
• Multi-threading support is limited to what C supports.
• Coroutines are not supported; however, custom for loops are supported that work like macros.
• goto and labels are not supported.
• String interpolation is not supported to simplify the language. Instead, use an arrays of strings. As commas are optional, this is short.
• Unsigned integer are intentionally not supported to simplify learning and using the language, to avoid surprising behavior and edge cases. Unsigned integer are specially tricky if combined with type inference, as in this language. Mixing signed and unsigned values can result in unexpected behavior, and unsigned arithmetic can mask errors by turning invalid states into large valid-looking values. Some example are: int n = 100; for (unsigned i = n - 1; i >= 0; i--) { printf("%d\n", i); } and int a = -1; unsigned int b = 1; if (a < b) printf("a is less than b\n"); else printf("b is less than a\n");. When needed, unsigned behavior is available through explicit operations (eg. unsigned division). This design does not affect performance or memory usage.

Syntax

• Spaces (indentation) is used to group statements. This reduces the number of lines. Tabs are not supported. The reason is that spaces are more common, and tabs do not mix well with spaces. By disallowing tabs, problems are detected early.
• Commas in parameter lists are optional, if parameters are simple values. The same as in shell scripts or Lisp. This also makes 'print' statements more readable (without string interpolation).
• There is no boolean data type to simplify the syntax. Instead, true is 1 and false is 0. The common pitfalls, e.g. comparing the result of a comparison, requires parenthesis (eg. a > b < c is not allowed).
• Constants and variables are defined in a different way (: vs :=) so that it's easier to see for a reader if it may change later. But there is no keyword like "var", "val", "const", or "final" to shorten the code.
• Definition of a variables is distinct from updating it (:= vs =) to quickly detect if a variable was already defined, and to detect typos.
• break and continue can have a condition, to avoid a separate line with if.
• Labels for break and continue are not supported to simplify the language. If needed, the function can return from inside the loop, or throw an exception (such exceptions are very fast).
• Comments are only a single character (#) to save some typing. Block comments (##) are useful if the editor doesn't support commenting a block. To support eg. Markdown inside of block comments, the delimiters can be variable length.
• Raw strings are useful, to avoid escaping problems: https://xkcd.com/1638/
• Multi-line strings are always raw strings, as escape sequences don't seem useful for this (tabs are supported here).
• Dangling , are supported to e.g. simplify re-ordering entries.
• Bit operations |, &, ^, ~ have a higher order of precedence than comparison. This is different from other programming languages. It seems the reason why it is different in other languages is historical reasons only.
• Instead of && || ! we use the keywords and, or and not, to make these common cases easier to understand for new developers, and in case of not to make it easier to read.
• There are no separate unsigned data types, to simplify the language.
• Bitwise shift to the right (>>) is a logical shift, that means for negative values, a number of zeros are added to the left. The arithmetic shift is not supported by the language itself, but can be supported by a library function (same as eg. rotation). The reason is that logical shifts are more common.
• A main method is not needed. However, if there is no main method, then the variables before the first function (if any) are global variables, and those after a function are local variables.

Memory Management

• Reference counting is used for reference types.
• Mark-and-sweep garbage collection is not used to avoid pauses.
• Borrow checking is not used to simplify writing code.
• The plan is to use reference counting only where cycles are not possible.
• The plan is to support weak references.
• The plan is to support unique pointers, and arrays of pre-allocated objects accessed via handlers and a generation

Safety

• This language is memory-save. There is no way to write memory unsafe code, except by calling C methods.

Array Bound Checks

• Where array bound checks are needed, and the index is out of bounds, the program panics.
• Array bounds are check, except if array access is guaranteed to be inside the bounds. This is implemented using dependent types. Internally, this language uses a symbolic reasoning engine for linear integer arithmetic, which is a bit more complex that what most compilers use, but not as complex (and powerful) as eg. Ada SPARK.

### Null Pointer Access

• Possible null references need to be handled. This is checked at compile time. There is no way that null references can throw an exception or panic.

Division and Remainder (Modulo) By Zero

• Potential integer division (/) by zero is detected at compile time and not allowed.

Exceptions and Panic

• Exceptions need to be handled using catch, or re-thrown.
• There is no try keyword: catch will catch all exceptions in the same scope. This is to simplify the code, and reduce the need of indentation. Ruby supports a similar syntax: begin is not needed.
• Custom exception types are allowed, with some restrictions: Exception types need to have an integer field exceptionType that may not have a negative value (because internally, this field is used to to flag whether the method was successful or not, and a negative value is used to indicate success).
• If a type has a close() function, it is called when the memory is freed. If this function re-adds a reference to the object, then the program panics.