Adding reset_data to xbuffer_adaptor and reset_buffer to *adaptor to replace the pointer without any reallocation.

[tdegeus]

Checklist

• The title and commit message(s) are descriptive.
• Small commits made to fix your PR have been squashed to avoid history pollution.
• Tests have been added for new features or bug fixes.
• API of new functions and classes are documented.

Description

Fixes #2519

[tdegeus]

Grateful for your feedback @emmenlau . I'm not sure that I like the name of the method, happy to take suggestions.

[tdegeus]

Restarting CI

[emmenlau]

Thanks a lot for the nice work @tdegeus ! Could you also take a look, @vakokako ?

About the name of the method, I find it also difficult to come up with a good suggestion. What about setdata() or setptr()? Or just ptr() if it can be used as a setter and getter? Just tossing ideas...

[vakokako]

Looks good. The problem is it's not really discoverable with xbuffer_storage and I had to dive into the code to really find this method. Also would be still nice to have this functionality implemented in a view as the intended purpose of the xt::adapt is wrapping non-xtensor containers rather than providing a movable view onto the existing xtensor container.

[tdegeus]

Thanks for your feedback @vakokako .

The problem is it's not really discoverable with xbuffer_storage and I had to dive into the code to really find this method.

Do you mean documentation is lacking? Do you think that this PR sufficiently helps in that direction?

Also would be still nice to have this functionality implemented in a view as the intended purpose of the xt::adapt is wrapping non-xtensor containers rather than providing a movable view onto the existing xtensor container.

I see your point. Though for a view I don't really have a good syntax in mind. Anyway it's a separate enhancement I guess. Would you be willing to open an issue on that, with a suggested API?

[vakokako]

Do you mean documentation is lacking? Do you think that this PR sufficiently helps in that direction?

It's just hidden under the .storage() method, while it could be the direct method of the adapted container like so:

auto b = xt::adapt(&a.flat(0), std::array<size_t, 2>{2, 2});
b.replace(&a.flat(1)); // maybe could be called `move`?

The types in xtensor are already hidden enough, without diving into code I wouldn't even know what type storage() is, that it has also some size, etc.

But I think this is a minor remark, thanks for implementing this pr!

[tdegeus]

@JohanMabille Can you check that the edit is what you had in mind?

[JohanMabille]

Awesome, thanks!