Sebastian Berg
Sebastian Berg
Going to close this since I forgot to tag the PR. Nested CAI will now be supported (v14); i.e. nesting an object should be supported if it is supported by...
@jbrockmendel yes, but it is completely removed in 2.2 (or 2.3?).
Sorry, I guess the point is: Yeah, I think you can just close it. The promotion state was never pefect in 1.26 and the "weak and warn" was a best...
No, it can't be changed. The intention of the promotion state was for testing: both to prepare libraries for the switch and to see how the change affected downstream. Since...
> there is supposed to be a per-module import lock that is held by the interpreter during import - doesn't that protect us here? Randomly saw this part and it...
I think this has been discussed in cupynumeric, but not sure of the current status/priority. I'll bring it up with the cupynumeric team. Not sure what is best, in practice...
The problem is not the overflow, it's the `float16()`. That is a typical issue with NumPy that it picks the lowest loop (which is just not really the right thing)....
I don't *think* this was ever regression though. Even in NumPy 1.x the same thing shoudl have happened, so this behavior was around forever.
Ah, right. You should see the same with `np.ldexp(1, np.array([16]))` where the second arg is not 0-D. But that is probably pretty uncommon even if `math.ldexp` is probably often nicer...
This behavior is explained in the documentation notes. You probably want to use `indexing="ij"` which is more regular/obvious in its behavior. (Is that the better default? Maybe not, the history...