mirror of
https://github.com/angr/angr
synced 2026-08-17 12:23:11 -04:00
d4a3f058e0
Since Python 3.12, builtin sum() uses Neumaier compensated summation for floats, which is more accurate than the program's naive left-to-right addition. The test constrains the program's sequential FP sum to equal 27.7 exactly, then asserted sum(args) == 27.7 — which additionally demands that the model have no intermediate rounding error under compensated summation, something the constraint does not guarantee. Sum with functools.reduce to mirror the program's sequential addition. Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com> |
||
|---|---|---|
| .. | ||
| block | ||
| __init__.py | ||
| test_argc.py | ||
| test_argc_sym.py | ||
| test_argv.py | ||
| test_callable.py | ||
| test_windows_args.py | ||