Wessel

Hey @jlapeyre, This is a good suggestion! I'm fully on board with this. I agree that having examples/answers to the bullet points would be really nice. I'm going to be...

I also know of [Geometric Kernels](https://github.com/GPflow/GeometricKernels), which uses Plum to provide a backend-agnostic implementation (something that works with PyTorch/TF/JAX/etc). There is [fastdispatch](https://github.com/fastai/fastdispatch), which I believe is eventually to be used...

Hey @PhilipVinc, Sorry for not getting back to this earlier. I'm a bit busy at the moment. I fully agree that being able to dispatch on the dimensionality of an...

I've made the above change in the most recent release, `1.1.1`, and added [an entry to the README](https://github.com/wesselb/plum#hooking-into-type-inference). I'll leave this issue open and close it once good implementations of...

Hey @PhilipVinc and @EelcoHoogendoorn, Ah @PhilipVinc says, I think this might be challenging, but it could conceivably be done. I've hacked something very quick together. Would something like the below...

> The ability to specify some kind of computable matching function is quite critical for my actual application though. Would something like how the example works suffice, or would you...

Hmm, perhaps I'm misunderstanding, but I think what you're after is possible in the setup of the example. E.g., what about something like the following? ```python from plum import dispatch,...

> With the minor detail that I think usually itd be nice to seperately bind the attribute to create a len/whatever_dispatcher since that part tends to be constant over a...

That's totally fair enough. > getting exactly the syntax I want takes me 40 lines of code added to my project. How would you deal with the below ambiguity? ```python...

I see! If that approach or assigning a priority number suffices, then perhaps it is not necessary to go through a lot of technical hoops for a fully general implementation...