pandas Windows vs Unix: Calling Index.__init__ with dtype='int' yields either RangeIndex or Int64Index

Question

I have observed a difference in running pandas on Windows vs Unix.

I create a fresh Python 3.9.12 environment on both Windows and Unix and install pandas 1.3.4 in both. I observe the following

---

Input:

pd.Index(range(3), dtype='int')

Output on Windows:

Int64Index([0, 1, 2], dtype='int64')

Output on Unix:

RangeIndex(start=0, stop=3, step=1)

Summary: Windows and Unix disagree for dtype='int'.

---

Input:

pd.Index(range(3), dtype='int32')

Output on Windows:

Int64Index([0, 1, 2], dtype='int64')

Output on Unix:

Int64Index([0, 1, 2], dtype='int64')

Summary: Both Windows and Unix return Int64Index for dtype='int32'.

---

Input:

pd.Index(range(3), dtype='int64')

Output on Windows:

RangeIndex(start=0, stop=3, step=1)

Output on Unix:

RangeIndex(start=0, stop=3, step=1)

Summary: Both Windows and Unix return RangeIndex for dtype='int64'.

---

My question: Any advice on what to write? I prefer consistent behavior across Windows and Unix. I guess I should refrain from (ever?) writing 'int', so should I instead write 'int32' or 'int64'?

This discussion relates: Why do Pandas integer `dtypes` not behave the same on Unix and Windows?