Closed qTich closed 10 months ago
The magic number from renpy-8.2.0.23121103+nightly-sdk-ci/lib/python3.9/jnius/env.pyc
is 61 0d 0d 0a
.
The magic number from the python3.9 binary that ships with the same Ren'Py SDK matches:
>>> print(*('{:02x}'.format(x) for x in importlib.util.MAGIC_NUMBER))
61 0d 0d 0a
The a7 0d 0d 0a
magic number (aka b'\xa7\r\r\n'
from OP) is the magic number of py3.11.
Given the above, it seems unlikely that this behaviour was observed by building from a nightly SDK (if I'm incorrect in this, please provide the exact version). If resulting from your own builds, is it possible that you built the web version prior to building the android version?
The magic number from
renpy-8.2.0.23121103+nightly-sdk-ci/lib/python3.9/jnius/env.pyc
is61 0d 0d 0a
. The magic number from the python3.9 binary that ships with the same Ren'Py SDK matches:>>> print(*('{:02x}'.format(x) for x in importlib.util.MAGIC_NUMBER)) 61 0d 0d 0a
The
a7 0d 0d 0a
magic number (akab'\xa7\r\r\n'
from OP) is the magic number of py3.11.Given the above, it seems unlikely that this behaviour was observed by building from a nightly SDK (if I'm incorrect in this, please provide the exact version). If resulting from your own builds, is it possible that you built the web version prior to building the android version?
I think it's enough to build The Question
on one of the last two nightly builds to see the problem.
Broken .pyc:
'/lib/python3.9/steamapi.pyc': b'\xa7\r\r\n',
'/lib/python3.9/jnius/signatures.pyc': b'\xa7\r\r\n',
'/lib/python3.9/jnius/reflect.pyc': b'\xa7\r\r\n'
Bytecode generated for the web version was leaking (pseudo-randomly due to glob order), a more targeted filter should ensure we always get the correct files.
adb8b029db5a4623fb355e53e8cc3b2f9222e4ac - breaks launching on android, i guess it affects all 3 libraries -
pyjnius
,pyobjus
,steam