Name of the "Base" package

[bgroenks96]

This repository is intended to host the long-awaited common abstractions and utilities for all NumericalEarth projects, as well as SpeedyWeather and Oceananigans.

We first need to decide on a name. A few different ideas have been floated, such as NumericalEarthCore, NumericalCore, NumericalEarthBase, and NumericalEarthAbstractions.

I would argue against the *Core names, not because they don't sound cool (they do) but because

• The convention set by Julia itself is that Base provides the extensible, public-facing functionality that packages and applications can build on, whereas Core is, AFAIK, entirely internal and not meant to be used. So the name NumericalEarthCore would be very explicitly breaking from this convention. We can still do it, of course, but that's something to be aware of.
• NumericalCore would be actively misleading because it implies that there are actual numerics in the package (like a "dynamical core") which is not what we plan to do. We just want a place put common names for methods and interfaces.

I guess @glwagner will have a different opinion.

CC @giordano @navidcy @simone-silvestri @maximilian-gelbrecht @milankl

[lazarusA]

EarthBase

[bgroenks96]

If anyone else would like to vote for this name, please upvote the comment.

[glwagner]

I preferred NumericalEarthCore just because it’s more poetic and interesting. But seems like it’s not preferred by others.

[simone-silvestri]

Ok then what about using something really explicative like NumericlaEarthSharedMethods or just NumericalEarthFunctions? In the end this package is going to be a "non-package" just a container for function names, so a very explicative name might work well without needing to convey more than the package is.

[glwagner]

Is it really a "core" module if all of the packages have existed up until now without it and it doesn't provide any actual functionality?

Yes, it is. "Core" implies something central and small. "Base" implies something large, a foundation to build upon. What is being proposed is a core, not a base.

[milankl]

I don't think I agree with either of you. "Core" is small yes, but also indispensable as its literally the core. I see that JuliaBase for example is indeed a large foundation, but that's also because the julia repo itself is huge. In SpeedyWeather we went for *Internals for a package that all other in the monorepo depend on without it being seen as "core" or "base" functionality. In that sense I opt for *Abstractions because I feel that both functions (their abstract names and not their specific methods) and abstract super types can be seen as abstractions.

[glwagner]

Maybe I am mistaken about the purpose of the package. It will only provide functions that are peripheral somehow rather than indispensable or central? In that case I agree, "core" is a misnomer.

[bgroenks96]

I guess they are central because functions like initialize! and run! will be used everywhere by all NumericalEarth packages.

But they are also just names.

[milankl]

Not against *Abstractions but *Base sounds more inclusive, as in abstractions could live in Base but there might be other things we want in Base that aren’t abstractions?