No way. It won't be responsive unless the load is actually something enough to discharge it quick.

Say you step from 0A to 1.5A suddenly. What will the rail do?

The LM317 has a very poor load transient response with just 1uF; it's rated for about -0.5V when the load increases, and +0.5V when it increases. (Page 8 of the datasheet.) Increasing the cap provides a "buffer" which keeps the output voltage stable during transient loads. Remember, a lab power supply is intended to be as perfect a voltage source as possible.

The supply schematic I drew up has very good transient response of less than 5mV down and 2mV up.

 

A common bench mark for a decent supply is 5mV ripple PP, and +/- 0.1% regulation.

You won't get this with a switcher.

I spent about the last 15 years doing that with switchers. It is easy to get the ripple that low if you want to spend the money for the filter components and you know how to manage the mechanical design to isolate the switching noise. In a lot of cases, the design doesn't require it, so they don't spend all the $$$ it takes to get it that low.

 

Say you step from 0A to 1.5A suddenly. What will the rail do?

The LM317 has a very poor load transient response with just 1uF; it's rated for about -0.5V when the load increases, and +0.5V when it increases. (Page 8 of the datasheet.) Increasing the cap provides a "buffer" which keeps the output voltage stable during transient loads. Remember, a lab power supply is intended to be as perfect a voltage source as possible.

A good rule of thumb on linear supplies is to ratio the input/output filter cap sizes at about 3/1 to support load transients. If the input filter cap is 10,000uF go with 3 - 4,000uF on the output. No matter how fast the loop response of the regulator is, you still need output capacitance to support fast load transients.

 

I spent about the last 15 years doing that with switchers. It is easy to get the ripple that low if you want to spend the money for the filter components and you know how to manage the mechanical design to isolate the switching noise. In a lot of cases, the design doesn't require it, so they don't spend all the $$$ it takes to get it that low.

Absolutely right. Switchers can be very low noise. I've been working with a micro buck converter. 4.5V - 20V in, 3.3V out @ 1A, ripple <5mVp-p with just a 22µF ceramic cap.

My HP 54501A (oscilloscope) has a very old switching power supply in it. The input channel buffer uses the +12V and -12V rails from the power supply directly (no regulator in sight.) And yet its noise is <1mVp-p. The actual noise from the power supply is very difficult to measure with my second analog scope, it's within the noise floor of about 500µV! It has about 8800µF per rail of capacitance.