Let's say you have to do a parallel or tear-drop entry to enter this distance-based holding pattern. Do you still fly outbound from the fix for one minute and then make your turn inbound to re-intercept the fix or are you taking it to 4 NM?
I haven't found any specific guidance for hand flying entries to RNAV (RNP) holding patterns. But part of the concept of RNAV is that the navigation system can provide the complete desired path including hold entries.
Referring to RTCA DO-236C RNP MASPS and the requirements for holding operations, I find the following: The dimensions of the hold pattern are defined based on altitude, airspeed, bank angle, and an allowance for wind. The length of the inbound/outbound leg is defined by the database or entry. In your case this is 4 NM, and is defined as d1. The width of the pattern (d2) is computed based on the above variables.
Without the details, we can still make an approximation. If you're flying at 120 kts and performing a 2 min turn (1 min for 180 deg), your pattern width (d2) would be about 1.3 NM.
The rules include the following statement:
When performing an entry into the holding pattern, the navigation system is permitted to use an inbound leg length (d1) that is equal to the holding pattern width (d2) instead of the length defined in the database or from pilot entry.
So the end result is that your entry cannot exceed the 4 NM leg length, and could have an inbound leg length as short as 1.3 NM.
It is possible to reach the holding fix and enter the hold using any of the three entries described in the AIM, fly outbound on a specific heading for a given length of time, and then turn inbound and intercept the inbound course exactly at the point which is the required along track distance from the fix, or required inbound time to the fix.