Bugfix - calls to private functions with multiple lists as parameters

[iamdefinitelyahuman]

What I did

Fixed an bug stemming from #1470 - calls to private functions that had a bytes list following an integer list of length > 1 were underflowing.

In the following example calls to bar would underflow, but bar2 would succeed:

@private
def _foo(y: int128[2], x: bytes[5]) -> int128:
    return y[0]

@private
def _foo2(x: bytes[5], y: int128[2]) -> int128:
    return y[0]

@public
def bar() -> int128:
    return self._foo([1, 2], b"hello")

@public
def bar2() -> int128:
    return self._foo2(b"hello", [1, 2])

How I did it

Used get_static_size_of_type to verify the memory size of each argument during iteration.

How to verify it

I have included a test as the first commit, if you run it without the 2nd commit you can see the bug in action.

Description for the changelog

• Bugfix - multiple arrays as inputs in a private function

Cute Animal Picture

[fubuloubu]

Only python 3.6 build failed, with the following message:

E           hypothesis.errors.Flaky: Hypothesis test_sqrt_valid_range(sqrt_contract=<tests.base_conftest.VyperContract at 0x7fea210e60b8>, value=Decimal('1760211996687216922475948105605887618.3833309761')) produces unreliable results: Falsified on the first call but did not on a subsequent one

Retriggering to see if results repeat.

[iamdefinitelyahuman]

@fubuloubu I get a similar fail output on that test about 20% of the time when I run tests locally, including on master 🤷‍♂️ From the output it seems it's just failing because the test takes too long to finish, so I've been ignoring it.

[charles-cooper]

@fubuloubu I get a similar fail output on that test about 20% of the time when I run tests locally, including on master man_shrugging From the output it seems it's just failing because the test takes too long to finish, so I've been ignoring it.

I think I've run into a similar bug. I think we need to increase the timeout on that test to like 1s (currently 400ms I believe).

[charles-cooper]

nice catch. thanks

[charles-cooper]

The change LGTM. I found the test case confusing though, I think it will be clearer if you just use the test from the PR description - unless this tests something that is not in the PR description and I am missing something. Thanks.

[iamdefinitelyahuman]

My goal in the test was to ensure I had many lists with length > 1, and started one with a bytes list and another with a not-bytes list. In hindsight yes, it's maybe a bit overkill. Will update.

[iamdefinitelyahuman]

@charles-cooper i've revised the test per your comments

[charles-cooper]

LGTM - thanks!

[charles-cooper]

@jacqueswww can I merge this?

[jacqueswww]

LGTM