How is data stored in old Go maps after defining a bucket?

In older implementations of Go maps (before Go 1.9), data was stored in a structure where keys and values were grouped into buckets. Each bucket contained a fixed number of key-value pairs.

After determining the bucket for a key (using a hash function and a mask), all pairs with hashes falling into that bucket were stored in an array within the bucket. If the bucket overflowed, the map was resized (rehash) with an increased number of buckets.

Thus, data within a bucket was stored as arrays of keys and values, which allowed efficient searching and inserting of elements, minimizing collisions.

This differs from newer implementations, which use a more complex structure with optimizations to reduce copying and speed up access.