The NASM translation should probably note in the details that count can be zero.

Thanks. That worked. Just a suggestion, maybe put that in the description? Maybe it's just me but it doesn't really seem obvious, especially when you're trying to be efficient.

Python random tests seem to leave the first value off the expected output.

Input: [-63, -62, -60, -58, -56, -55, -53, -52, -51, -50, -47, -45, -44, -43, -40, -39, -36]
Error: It should work for random inputs too: '-63,-62,-60,-58,-56,-55,-53--50,-47,-45--43,-40,-39,-36' should equal '-62,-60,-58,-56,-55,-53--50,-47,-45--43,-40,-39,-36'

Input: [-91, -88, -86, -85, -82, -81, -80, -78, -75, -74, -72, -71, -70, -67, -64, -62, -60, -57, -55, -54, -53, -52, -50]
Error: It should work for random inputs too: '-91,-88,-86,-85,-82--80,-78,-75,-74,-72--70,-67,-64,-62,-60,-57,-55--52,-50' should equal '-88,-86,-85,-82--80,-78,-75,-74,-72--70,-67,-64,-62,-60,-57,-55--52,-50'

Input: [-61, -59, -56, -55, -53, -50, -47, -44, -42, -40, -38]
Error: It should work for random inputs too: '-61,-59,-56,-55,-53,-50,-47,-44,-42,-40,-38' should equal '-59,-56,-55,-53,-50,-47,-44,-42,-40,-38'

I look forward to this kata when it is translated into Python or C. For now, I wait.

Created a translation for Python. Tell me if there's anything I need to fix.