MAINT: random: Match type of SeedSequence.pool_size to DEFAULT_POOL_SIZE. #14303
Conversation
…IZE.
Fixes this compiler warning:
gcc: numpy/random/bit_generator.c
numpy/random/bit_generator.c:4889:41: warning: comparison of integers of different signs: 'int' and 'uint32_t' (aka 'unsigned int') [-Wsign-compare]
__pyx_t_6 = ((__pyx_v_self->pool_size != __pyx_v_5numpy_6random_13bit_generator_DEFAULT_POOL_SIZE) != 0);
~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1 warning generated
| @@ -14,7 +14,7 @@ cdef class BitGenerator(): | |||
| cdef class SeedSequence(): | |||
| cdef readonly object entropy | |||
| cdef readonly tuple spawn_key | |||
| cdef readonly int pool_size | |||
| cdef readonly uint32_t pool_size | |||
There was a problem hiding this comment.
Is there a reason this (and the other counts in this struct) isn't size_t?
There was a problem hiding this comment.
pool_size is supposed to be tiny (relatively speaking). n_children_spawned could be largish, but should be capped at uint32_t (as it turns into a uint32_t element in the spawn_key).
There was a problem hiding this comment.
The docstring for the pool_size attribute says "Size of the pooled entropy to store. Default is 4 to give a 128-bit entropy pool. 8 (for 256 bits) is another reasonable choice if working with larger PRNGs, but there is very little to be gained by selecting another value." So 32 bits is far more than necessary, but I guess it is harmless to leave it that size.
n_children_spawnedcould be largish, but should be capped atuint32_t(as it turns into auint32_telement in thespawn_key).
OK, I changed n_children_spawned to uint32_t in the changes that I just pushed.
|
In the latest update, both |
| @@ -116,7 +116,7 @@ def _coerce_to_uint32_array(x): | |||
| Examples | |||
| -------- | |||
| >>> import numpy as np | |||
| >>> from np.random.bit_generator import _coerce_to_uint32_array | |||
| >>> from numpy.random.bit_generator import _coerce_to_uint32_array | |||
There was a problem hiding this comment.
Why is this needed? The line above should make the old code work ...
There was a problem hiding this comment.
Why is this needed?
I'm not sure which part of the change "this" refers to. The import statement is needed so we can use _coerce_to_uint32_array without giving the fully qualified name. The change from np to numpy is needed because a Python variable can't be used as part of the module in an import statement.
There was a problem hiding this comment.
I don't think the import numpy as np is needed for examples, but that is extraneous to this PR.
|
@WarrenWeckesser Some lines in your commit message are way long. Might want to check your editor setting for them. Most editors have plugins for git commit messages. |
|
@charris, the long lines are from the compiler warning that I included in the commit message. I intentionally didn't break that up, but I can do so in the future. |
Fixes this compiler warning:
gcc: numpy/random/bit_generator.c
numpy/random/bit_generator.c:4889:41: warning: comparison of integers of different signs: 'int' and 'uint32_t' (aka 'unsigned int') [-Wsign-compare]
__pyx_t_6 = ((__pyx_v_self->pool_size != __pyx_v_5numpy_6random_13bit_generator_DEFAULT_POOL_SIZE) != 0);
~~~~~~~~~~~~~~~~~~~~~~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1 warning generated