ENH: Support arbitrary-length Python ints in QuadPrecision constructor #213
Conversation
|
I'd like to give the C code a careful once-over later today. |
Cool, I'll do maybe 2 more PRs today (its a holiday here so xD) |
| else if (lval == -1 && PyErr_Occurred()) { | ||
| Py_DECREF(py_int); | ||
| Py_DECREF(self); | ||
| return NULL; |
There was a problem hiding this comment.
you can test this error case by defining a class with an __index__ implementation that raises.
There was a problem hiding this comment.
Somehow this isn't failing it, maybe need more low level of int corruption to trigger this.
Anyways rest are added
|
Thanks @ngoldbaum mostly reviews seem to be connecting to more exhaustive testing, I'll soon add the tests to address them |
|
@ngoldbaum gentle reminder here, if this is GTG and also the #221 |
|
Looks like we had a race condition. I just submitted the review for #221 and was going to look at this next. |
3 participants
This PR is a part of work, where making
numpy_quaddtypecompatible with NumPy's longdouble testsDetails
Use
PyLong_AsLongLongAndOverflowinstead ofPyLong_AsLongLong, and handle the overflow case explicitly by reconstructing theQuadPrecisionobject via string conversion.