Name of the game

[mmaechler]

I think we should spend some effort now to really change the name away from R7 ASAP.

This is not about R versions, nor is it about summing S3 + S4, ..

I agree a short 2-letter name may be very desirable, but starting it with R may be suboptimal, unless we use a non-digit as 2nd letter, e.g., RO

[mmaechler]

I mentioned O1 (The number one definitive object system) when we met online recently ..
(note this is "O 1" (without the space), and above "R O" (2 capital letters), the backquote font here really showing the letter O almost as the digit 0 (O vs 0).

[hadley]

To remind ourselves of the thinking behind R7:

• R: this was an OOP system designed for specifically for R, as opposed to S
• 7: 3 and 4 have strong associations with S3 and S4, R5 was used unofficially to refer to reference classes, and R6 is an existing package. 7 also has the nice implication of 3 + 4.

In my mind, the major problem with R7 is that it implies a strong connection with R6, i.e. it makes people think that R7 is an OOP system where methods belong to classes, not functions.

[mmaechler]

Yes, R5 was strongly disliked by the creator of the reference classes but inspite gave rise to naming 'R6' and indeed our "O1" is far from the R6 package.
S is basically unknown to 95% of the active R developers and users and "specifically for R, .. not S" does not really make sense to them.

[lawremi]

I'd like to broaden our search beyond names consisting of a letter and a number.

[hadley]

@lawremi have you considered everything in outer(LETTERS, 0:9, paste0) ?

[mikmart]

Would S7 make any sense?

As someone who just recently found out about this initiative, R7 did indeed make me instantly think OOP and immediately afterward "better R6." I think S7 would still preserve that strong association to OOP via S3 and S4, and also hint it's not continuing along the lines of R6 style OOP but instead the S3/4 style.

[lawremi]

Good point, S7 seems at least as reasonable as R7 to me.

[hadley]

I could get behind S7.

[mmaechler]

As mentioned yesterday at the virtual meeting, I think S7 is better than anything else mentioned till now. It has at least two clear advantages over R7 : S3+S4 = S7 \ not confusible with a future version of R.

[jimhester]

S is basically unknown to 95% of the active R developers

This seems a compelling argument to avoid using S in the name.

I think something simple like just oop or even oo might work.

For fun I asked everyone's favorite AI ChatGPT for some suggestions, here is what it came up with (verbatim including the rational for each)

• roop (short for "object-oriented programming")
• rplus (suggesting an improved version of R)
• roopr (combining the "R" in "R7" with the "oop" in "object-oriented programming")
• rclassy (suggesting a more elegant and sophisticated approach to OOP)
• rtoolbox (suggesting a collection of useful tools for OOP)
• rkit (short and simple, with a playful connotation)
• roopster (suggesting a more powerful and versatile OOP system)

[Robinlovelace]

👍 4 S7 interested to hear what others think...

[mmaechler]

I think starting with an "R" (lower- our upper case) is not a particularly good idea.
Using "oop" (one way or the other) was mentioned by Michael @lawremi early, too; I could live with a version of that and would prefer it to R7 ... then, given the proposals so far, my personal favorite would remain with S7
{One drawback of a version of "OOP" is that too many people mentally directly jump to C++ / Java style of OOP when they see/hear it}.

[s-fleck]

S7 (and S5) would imho be good choices. The R names are now established for reference-class style classes.

[tdeenes]

S is basically unknown to 95% of the active R developers

This seems a compelling argument to avoid using S in the name.

I am sure S3 and/or S4 are known to at least 95% of the active R developers (not users). For such a core package as this pkg will evolve I would definitely avoid ChatGPT-like names, which is basically a pseudoword generator in English with the only constraint "start with letter r" for this exercise. S7 is short, easy to remember, language-agnostic, and has an inherent relationship to the main focus of this pkg.

[t-kalinowski]

I like S5. It has continuity, and suggests that this is a further iteration on the ideas along the path of S3 --> S4 --> S5.

[DrEntropy]

Perhaps it should have a name that harmonizes with the tidyverse? methodr ? genericr ?

[mmaechler]

Perhaps it should have a name that harmonizes with the tidyverse? methodr ? genericr ?

Not really. The whole is not related to the tidyverse (apart from the fact that partly the same people are involved);
rather more to "all of R".

Could we rather try to finalize and start putting up a short list from which to choose?
I'd start with (alphabetical ordering) {R7, S5, S7}

[hadley]

In the OOP WG meeting on Monday there was broad agreement that S7 was the way forward.

[mmaechler]

Thanks, good to know (that meeting was not in my calendar for some reason). So this will be closed together with the name change (which will be done by ??; I could if wanted, but maybe you need to coordinate this with more than just this repos?)

[mmaechler]

That would be great. I have a somewhat interesting plan for a new R package to be based on 'S7',
current tentative name 'arbR' (interface of R and 'arblib' aka 'arb' --> https://arblib.org ) -> using S7 classed intervals (number +- errorbound).

[jl5000]

I don't know whether it's of interest but I've also jumped the gun a bit and am writing a fairly chunky package called gedcomR7 (name to change), which is working out great. I'm just having trouble trying to work out how to export @, methods, and friends (while S7 is a separate package).

[daattali]

The R7 -> S7 change needs to be updated in the GitHub repo's "About" section