Open mmckerns opened 2 years ago
Several KeyError
are thrown from the mutiprocess
tests for pypy-3.6
, pypy-3.7
, and pypy-3.8
-- due to semaphore/resource tracking issues. However, these same KeyError
are present when testing multiprocessing
... so I'm going to ignore them for now.
The multiprocessing
tests throw errors for pypy-3.8
(specifically, test_preload_resources
, test_timeout
, test_heap
, and several others due to "too many files open"). This is also seen with multiprocess
, however, by skipping the three named tests noted, the other failures ('too many files open') disappear. I'll leave this as an issue to be resolved by pypy
, and skip the three noted tests for now.
It seems that a TypeError
that hangs the tests is now thrown "infrequently". Given enough tries, the tests generate a TypeError
for pypy-3.6
, pypy-3.7
, and pypy-3.8
. However, it's not always due to a UserDict
as seen above. Sometimes, the error is:
Traceback (most recent call last):
File "<string>", line 1, in <module>
File "/Users/mmckerns/src/pypy3.7-v7.3.7-osx64/site-packages/multiprocess-0.70.13.dev0-py3.7.egg/multiprocess/spawn.py", line 105, in spawn_main
exitcode = _main(fd)
File "/Users/mmckerns/src/pypy3.7-v7.3.7-osx64/site-packages/multiprocess-0.70.13.dev0-py3.7.egg/multiprocess/spawn.py", line 115, in _main
self = reduction.pickle.load(from_parent)
File "/Users/mmckerns/src/pypy3.7-v7.3.7-osx64/site-packages/dill-0.3.5.dev0-py3.7.egg/dill/_dill.py", line 332, in load
return Unpickler(file, ignore=ignore, **kwds).load()
File "/Users/mmckerns/src/pypy3.7-v7.3.7-osx64/site-packages/dill-0.3.5.dev0-py3.7.egg/dill/_dill.py", line 557, in load
obj = StockUnpickler.load(self)
File "/Users/mmckerns/src/pypy3.7-v7.3.7-osx64/lib-python/3/pickle.py", line 1101, in load
dispatch[key[0]](self)
File "/Users/mmckerns/src/pypy3.7-v7.3.7-osx64/lib-python/3/pickle.py", line 1449, in load_reduce
stack[-1] = func(*args)
TypeError: weak object has gone away
I'll limit the remainder of this issue to pypy3
infrequently throwing a TypeError
.
Note that the Traceback for both TypeError
experienced are identical... except for the TypeError:
message itself.
pypy3
produces severalKeyError
and, potentially, tries to hash aUserDict
(thus throwing an error) when run with aforkserver
or byspawn
.Errors are as follows:
and