On Tue, 2019-09-10 at 17:28 +0200, Hameer Abbasi wrote:
On 07.09.19 22:06, Sebastian Berg wrote:
On Fri, 2019-09-06 at 14:45 -0700, Ralf Gommers wrote:
<snip>
Let me try to move the discussion from the github issue here (this may not be the best place). ( https://github.com/numpy/numpy/issues/14441 which asked for easier creation functions together with `__array_function__`).
I think an important note mentioned here is how users interact with unumpy, vs. __array_function__. The former is an explicit opt-in, while the latter is implicit choice based on an `array-like` abstract base class and functional type based dispatching.
To quote NEP 18 on this: "The downsides are that this would require an explicit opt-in from all existing code, e.g., import numpy.api as np, and in the long term would result in the maintenance of two separate NumPy APIs. Also, many functions from numpy itself are already overloaded (but inadequately), so confusion about high vs. low level APIs in NumPy would still persist." (I do think this is a point we should not just ignore, `uarray` is a thin layer, but it has a big surface area)
Now there are things where explicit opt-in is obvious. And the FFT example is one of those, there is no way to implicitly choose another backend (except by just replacing it, i.e. monkeypatching) [1]. And right now I think these are _very_ different.
Now for the end-users choosing one array-like over another, seems nicer as an implicit mechanism (why should I not mix sparse, dask and numpy arrays!?). This is the promise `__array_function__` tries to make. Unless convinced otherwise, my guess is that most library authors would strive for implicit support (i.e. sklearn, skimage, scipy). You can, once you register the backend it becomes implicit, so all backends are tried until one succeeds. Unless you explicitly say "I do not want another backend" (only/coerce=True).
The thing here being "once you register the backend". Thus requiring at least in some form an explicit opt-in by the end user. Also, unless you use the with statement (with all the scoping rules attached), you cannot plug the coercion/creation hole left by `__array_function__`.
Circling back to creation and coercion. In a purely Object type system, these would be classmethods, I guess, but in NumPy and the <snip>
def my_library_func(array_like): exp = np.exp(array_like) idx = np.arange(len(exp)) return idx, exp
Would have all the information for implicit opt-in/Array-like support, but cannot do it right now. This is what I have been wondering, if uarray/unumpy, can in some way help me make this work (even _without_ the end user opting in). The reason is that simply, right now I am very clear on the need for this use case, but not sure about the need for end user opt in, since end users can just use dask.arange().
Sure, the end user can, but library authors cannot. And end users may want to easily port code to GPU or between back-ends, just as library authors might.
Yes, but library authors want to solve the particular thing above right now, and I am still not sure how uarray helps there. If it does, then only with a added complexity _and_ (at least currently) explicit end- user opt-in. Now, I am not a particularly good judge for these things, but I have been trying to figure out how things can improve with it and still I am tempted to say that uarray is a giant step in no particular direction at all. Of course it _can_ solve everything, but right now it seems like it might require a py2 -> py3 like transition. And even then it is so powerful, that it probably comes with its own bunch of issues (such as far away side effects due to scoping of with statements). Best, Sebastian
Cheers,
Sebastian
[1] To be honest, I do think a lot of the "issues" around monkeypatching exists just as much with backend choosing, the main difference seems to me that a lot of that: 1. monkeypatching was not done explicit (import mkl_fft; mkl_fft.monkeypatch_numpy())? 2. A backend system allows libaries to prefer one locally? (which I think is a big advantage)
[2] There are the options of adding `linspace_like` functions somewhere in a numpy submodule, or adding `linspace(..., array_type=np.ndarray)`, or simply inventing a new "protocl" (which is not really a protocol?), and make it `ndarray.__numpy_like_creation_functions__.arange()`.
Handling things like RandomState can get complicated here.
<snip>
_______________________________________________ NumPy-Discussion mailing list NumPy-Discussion@python.org https://mail.python.org/mailman/listinfo/numpy-discussion