rename: MathConst => Irrational

[StefanKarpinski]

I have to say this feels like a much better name:

julia> typeof(π)
Irrational{:π}

[quinnj]

+1

[tkelman]

I like. #11200 says hello though.

[StefanKarpinski]

Right, what we really need is a way to deprecate exported bindings. Bonus: don't autocomplete those or otherwise expose them as suggestions.

[waldyrious]

Should @math_const change its name too?

[StefanKarpinski]

Thanks for the feedback, @waldyrious!

[waldyrious]

Glad I can help :)

[dpsanders]

+1

[waldyrious]

Typo: Irationals is missing an "r". Also, I'm not sure I fully understand this comment. The beginning is redundant and the ending is kinda vague (especially since technically irrationals are reals).

[goszlanyi]

-1
Just an opinion: MathConst was a better name.

[simonbyrne]

Now we have generated functions, we could also get rid of the numeric val argument to the macro: since we're renaming, perhaps this would be a good time to make the switch?

[simonbyrne]

Also, I like the change: it helps make clear why pi != float(pi)

[ivarne]

I like this change, but 5 hours seems like a too short comment period, especially with a global audience.

[StefanKarpinski]

Yeah, sorry about that. Got a bit trigger happy.

[simleb]

I'm not convinced. It might look nicer to the eye but then why wouldn't π+1 be promoted to Irrational as well? What about √2?

What I'm saying is that using the word irrational, which has a well understood mathematical meaning, for a type that wraps mathematical constants which happen to be irrational but are just a tiny subset of all irrational numbers, especially when the language already has a Rational type with a widely different use (integers are promoted to rational for instance…), is confusing.

I know it's just a name, and the type of these mathematical constants is meant to be transient (in the sense that they almost always appear as part of expressions which convert them) but it is confusing I think.

tl;dr What is Irrational? The type used for mathematical constants… obviously!

[simonbyrne]

@simleb Good point. What about IrrationalConst?

[dpsanders]

+1 for IrrationalConst. (I was just typing almost the same.)

[ScottPJones]

👍 also for IrrationalConst

[goszlanyi]

We had a good, obvious and short name for this.
Why not leave it as is?

[StefanKarpinski]

MathConst was far from obvious – there was a fair amount of confusion about it over time and it always bothered me. To argue in favor of Irrational rather than IrrationalConst, this type is capable of representing arbitrary irrational values as long as you have a way to define them. It's not like anyone is going to come up with a clever encoding that can represent all irrational numbers.

[waldyrious]

I think IrrationalConst may be clearer since (in computing) using the term "constants" specifies that these are hand-picked numbers rather than a type meant to represent any irrational number.

[lostanlen]

typo here: "irriational" -> "irrational".

[simleb]

Well, the name Irrational is starting to grow on me. I agree with the arguments set forth by Stefan in #11929: Irrational simply couldn't mean anything else in the context of Julia. There is no way to even test if a number is irrational so there is no way to create them outside of how it is done now: by providing an algorithm to compute them to arbitrary precision. And since these numbers clearly have to be constant, there is no need for Const in the name. (Plus there is really no advantage using IrrationalConst over MathConst.)

Thus from now on I fully support the name Irrational for these mathematical constants (until someone writes a clever rebuttal, haha)!