`simple_continued_fraction`

[Tomaqa]

Please provide this information in the documentation, it is available only in the source code:

// Deal with non-uniqueness of continued fractions: [a0; a1, ..., an, 1] = a0; a1, ..., an + 1].
// The shorter representation is considered the canonical representation,

Also, I am disappointed that I cannot acces particular coefficients of the continued fraction, so the whole class is useless in the case one does not want to use the provided public functions. I want to use it for rational approxamations. So I also need to convert it to rational afterwards.

[NAThompson]

This seems like a reasonable ask, and not too difficult. Lemme think about how to communicate it.

[Tomaqa]

I added these member functions as a workaround:

    size_t size() const noexcept { return b_.size(); }

    const auto& operator [](int idx) const { return b_[idx]; }
    auto& operator [](int idx) { return b_[idx]; }

    void reserve(size_t size_) { b_.reserve(size_); }
    void push_back(Z c) { b_.push_back(c); }

Although the non-const functions may cause inconsistency with the private member variable x_, it seems that it is not used anywhere else than in the constructor (which suggests to actually remove the member variable and use only the parameter itself /or its local copy/).

Also, it is a question whether to check index bounds or not.