Changed max(start_idx, 2) * p to max(start_idx, p) * p. This is done so that we always start from the max. of the smallest multiple of p in the range [l, r] and p * p. It is more beneficial to start from p * p rather than 2 * p as lower multiples of p would have been covered by the smaller prime factors.

Updating my fork

The text read: "Here we have an implementation that counts the number of primes smaller to n using block sieving" while actually the code counts the number of primes smaller than or equal to n.

Updating my copy of emaxx-eng

Changed "arrays of with values values" to "arrays of values" in line 1 of this sub topic

Updating my own fork of emaxx-eng

Here are the details and possible reasons for the change:
* Changed (k-n-1) th shift to (k-(n-1))th shift. The former evaluates to -1th shift for k = n and is wrong.
* Changed cyclic shift to cyclic left shift. I always got confused while reading this since I tend to forget whether it was the left shift or the right shift of b. So, I decided to add it explicitly since others may also face the same problem in the future without this info.
* Changed index nos. of k from [n .. 2n-1] to [n-1 .. 2n-2]. Since c[n] == c[2n-1], it might make more sense to start from n-1 since after doing this we get shifts in the increasing order i.e., 0th shift, 1st shift, ... , (n-1)th shift.
* Added a note in case anyone wonders about the value at c[2n-1].