Conversions extended

[castelao]

Extending module conventions with missing functions.

[castelao]

@DocOtak, this is the PR that I mentioned. It is still a WIP.

[DocOtak]

Thanks for point me at this.

For the SA <-> SR type functions, it will need the saar data, which is big, see:
https://github.com/TEOS-10/GSW-C/blob/main/gsw_saar_data.h

I think as raw float data it will be about 2MB compiled, but haven't tried to measure it.

[castelao]

Thanks for point me at this.

For the SA <-> SR type functions, it will need the saar data, which is big, see: https://github.com/TEOS-10/GSW-C/blob/main/gsw_saar_data.h

I think as raw float data it will be about 2MB compiled, but haven't tried to measure it.

Oh, I see. Thanks, I forgot about that dependency. Since it is a core functionality of GSW, it should be included despite the size requirement. One way of doing it is keeping that table behind a new feature, so anyone that needs a slim library such as for embedded systems, or don't need at all those functionalities could compile without it. Would you like to add that in another PR?

[DocOtak]

I'll take a look at it. In C the saar function looks to be doing some sort of interpolation and I'll need to figure out what it is doing (or maybe implement from a paper describing it). I haven't looked at the Matlab version yet.

Is there a preferred way to have the data added embedded like the way C does it? So far I've only really looked at include_bytes. In this library, I didn't see anyplace external data was being used outside of tests, and nothing seem to be being embedded. Is there a "better way" or more rusty to include all the lookup table data?