Ahh, this is an even more interesting case then. I haven't tried this out, but it should work.
The pulse generation counter still uses the pause trigger with everything the same as I listed before. Instead of an edge counter, you'll want to use a position measurement task (linear), z index enabled, initial position and z index value = number of pulses to generate, z index phase "A low B low", distance per pulse = 1, decoding type = 2 pulse counting, and export the "Counter output event" and set the output behavior to toggle. Connect your re-trigger signal to the Z input on the linear position counter, connect the A input to GND (low) and connect the output of the pulse generation counter to the B input. On the pulse generation counter also configure an Arm Start Trigger (which can be used in conjunction with a start trigger) and set it's input terminal to your retrigger signal.
If I'm correct, this should perform your desired action. There is big difference between this and the regular retriggerable finite pulse train generation. If you receive a trigger signal before your finite pulse train has completed, the whole application will fail. You will no longer get pulses trains of the correct length and shape. Just keep that in mind as you're working on this application.
I hope this helps!
gus....