Deprecate prime number functions (isprime, primes, primesmask, factor)

[simonbyrne]

See #16357.

Current functionality has been moved to https://github.com/JuliaMath/Primes.jl

[tkelman]

Cool. This should wait until Primes.jl is registered before merging.

edit: and move the docs into the readme, and add a NEWS.md mention

[tkelman]

Remaining mentions:

contrib/BBEditTextWrangler-julia.plist:645:            <string>isprime</string>
doc/manual/arrays.rst:322:    julia> x[map(isprime, x)]
doc/stdlib/numbers.rst:458:.. function:: isprime(x::Integer) -> Bool
doc/stdlib/numbers.rst:466:       julia> isprime(3)
doc/stdlib/numbers.rst:469:.. function:: isprime(x::BigInt, [reps = 25]) -> Bool
doc/stdlib/numbers.rst:477:       julia> isprime(big(3))
contrib/BBEditTextWrangler-julia.plist:861:            <string>primes</string>
contrib/BBEditTextWrangler-julia.plist:862:            <string>primesmask</string>
doc/stdlib/numbers.rst:480:.. function:: primes([lo,] hi)
doc/stdlib/numbers.rst:486:.. function:: primesmask([lo,] hi)
doc/stdlib/numbers.rst:490:   Returns a prime sieve, as a ``BitArray``\ , of the positive integers (from ``lo``\ , if specified) up to ``hi``\ . Useful when working with either primes or composite numbers.
contrib/BBEditTextWrangler-julia.plist:435:            <string>factor</string>
doc/stdlib/math.rst:1362:.. function:: factor(n) -> Dict
doc/stdlib/math.rst:1366:   Compute the prime factorization of an integer ``n``\ . Returns
 a dictionary. The keys of the dictionary correspond to the factors, and hence are of the same type as ``n``\ . The value associated with each key indicates the number of times the factor appears in the factorization.
doc/stdlib/math.rst:1370:       julia> factor(100) # == 2*2*5*5

edit: oh, and base/sysimg.jl, of course

[simonbyrne]

The question is how best to handle deprecations of functions: should we do the same thing as #13897, where we throw an error (instead of depwarn)?

[tkelman]

I actually liked the way you had it where the old functions were just moved to being included from deprecated.jl. No errors that way, just warnings.

[simonbyrne]

Yeah, I wasn't sure what was the best way (I just copied what Combinatorics.jl did), which is why I kept it as a separate commit. We should figure out a consistent way to do this, especially as we're going to move more stuff out of Base.

The trouble is figuring out a way so that doing the correct thing, e.g. something like using Primes; isprime(7), doesn't throw a warning or error for deprecation, duplicate export or method overwriting.

[tkelman]

So which (if any?) of these commits is able to do that?

[simonbyrne]

Once we merge this, my plan is to add something like the following to Primes.jl:

if VERSION >= v"0.6-"
    export isprime, primes, primesmask, factor
elseif VERSION >= v"0.5-dev+XXXX"
    import Base: isprime, primes, primesmask, factor
end

Then using Primes shouldn't cause a problem on any version (after 0.4).

[tkelman]

using isdefined would probably be more accurate in capturing exactly when the deprecations get removed.

[simonbyrne]

Okay, so we could use:

if VERSION >= v"0.5-dev+XXXX"
    if isdefined(Base,:isprime)
        import Base: isprime, primes, primesmask, factor
    else
        export isprime, primes, primesmask, factor
    end
end

Either way, we still need the commit no for this change.

[tkelman]

Yep. So if we prefer the error-causing version, merge this as is?

[simonbyrne]

Yes, I think it should be good to go.