Refactor unicornlib to use pybind11 #5516
Conversation
|
Can we use nanobind instead? |
|
(We already use nanobind in pypcode) |
|
I chose pybind11 because it cleanly integrates with setuptools, allowing us to get rid of most of the custom logic in setup.py. Nanobind doesn't seem to offer that, and instead requires integrating cmake. |
ebbe77f to
cfab23b
Compare
20c7aca to
312e23c
Compare
Let's discuss this offline. Given that the long-term goal being switching to P-Code and the fact that pypcode already uses nanobind, I don't think relying on CMake is a worse idea than the current situation. Personally I do not see pybind11 removing the performance overhead any time soon, and nanobind offers solid runtime performance boosts over pybind11, which is critical for the unicorn engine. |
|
Also translating nanobind and pybind11 code in between is almost trivial. |
for more information, see https://pre-commit.ci
|
Corpus decompilation diffs can be found at angr/dec-snapshots@master...angr/angr_5516 |
No description provided.