Cerebro’s supported control flow

Conditionals

Cerebro currently supports basic conditionals and assignment operators under those conditionals. It transforms such assignment operators into a compound statement that erases the use of conditionals.

For example, if we have the following conditional:

if cond:
        x = 5
else:
        x = 10

Then, Cerebro transforms it into:

x = cond * 5 + (1-cond) * 10

Limitations:

• Condition can only have a single conditional operator.

• Currently, only assignment operators are supported in the body of the conditional.

For-Loops

Currently, Cerebro uses the SCALE-MAMBA @for_range construct when using for-loops. No extra work needs to be done on the developer’s end, one can just write:

for i in range(n)

and have that code be transformed into a representation the underlying framework understands.

Limitations: (which can be resolved by unrolling the loop)

• Assignments to variables outside the for-loop scope cannot be made.

Loop-Unrolling

Loop unrolling is currently a work-in progress, but it allows code within a for-loop to be unrolled into a series of assignment statements and function calls.

After enabling loop unrolling, the following example code:

x = 0
for i in range(2):
        x += i

would be transformed into:

x = 0
i = 0
x += i
i = 1
x += i

It allows more flexibility in what to code in for-loops for the developer.

Function Inlining

Function inlining is also a work in progress. It replaces function calls with the entire inlined version of the function call. For example, if we have the following code:

def f(y):
        x = 5
        print("Hello")
        return x + y

def g(x):
        return f(x)

g(3)

After function inlining, it would be transformed into:

g_x = 3
f_y = 3
f_x = 5
print("Hello")
f_ret = f_x + f_y
g_ret = f_ret

Limitations:

• Currently function inlining is a bit wonky dealing with OOP.