MCPcopy Create free account
hub / github.com/pydata/numexpr / NumExprReg

Class NumExprReg

numexpr3/numexpr_object.hpp:37–44  ·  view source on GitHub ↗

We can pack the tuples in Python and pass them in directly? General idea here is to not access the Python API directly after initial processing.

Source from the content-addressed store, hash-verified

35// General idea here is to not access the Python API directly after initial
36// processing.
37struct NumExprReg
38{
39 char *mem; // Pointer to array data for scalars and temps (npy_iter used for arrays)
40 char dchar; // numpy.dtype.char
41 npy_uint8 kind; // 1 = array, 2 = scalar, 4 = temp, 8 = return
42 npy_intp itemsize; // element size in bytes (was: memsizes)
43 npy_intp stride; // How many bytes until next element (was: memsteps)
44};
45
46// Presently the PyObject_HEAD macro expands into a single PyObject
47// When copying data we generally don't want to overwrite the head.

Callers

nothing calls this directly

Calls

no outgoing calls

Tested by

no test coverage detected

Used in the wild real call sites across dependent graphs

searching dependent graphs…