Athan

I should state that the strided array function data (i.e., `types.json`, `data.js`, `abs.functions.js`, etc) do not need to be split over multiple files; they can be included in the same...

Yes, that should be true. The goal would be to abstract away the underlying implementation (i.e., the loops), so bundles should be relatively small. This is unlike in C where...

Yes, that is correct. `abs` is a bit of an extreme case, given the number of accepted input and output types. For `sin`, for example, the number of accepted type...

...but you're right that there is fairly significant room for developer error when writing these interfaces (e.g., matching elements across the various meta data arrays).

Re: compressed form. I suppose one design goal toward which I biased the implementation was to make the JS implementation mirror the C implementation. In C, we need separate lists...

At the JS layer, I did include validation. See [here](https://github.com/stdlib-js/stdlib/blob/7fb13b1cdcb55a60bfa1a343932793439c716a85/lib/node_modules/%40stdlib/strided/dispatch/lib/main.js#L85). But this does not provide the sort of locality which would make identifying bugs easier.

Re: how common. Well, if we decide to go with the current implementation, it will be rolled out to at least `200` packages in `stdlib`. 😅

Re: introspection. Yes, this would be a straightforward addition. Can hang a method off the dispatch function (see [here](https://github.com/stdlib-js/stdlib/blob/7fb13b1cdcb55a60bfa1a343932793439c716a85/lib/node_modules/%40stdlib/strided/dispatch/lib/main.js#L131)) quite easily based on the provided `types` array. So this would...

You are correct! Not straightforward at all. NumPy provides two APIs: `may_share_memory` and `shares_memory`, with the former just performing bounds checks, and the latter using a much more complicated algo...

@Planeshifter I think the problem is not so much that the info is split across multiple files, but that the info is split across multiple data structures. Because the data...