Inconsistent number of digits in exponents when printing float numbers

Help!
1

I’m using: SWI-Prolog version 8.2.4 (pre-built Windows 64-bit version) (but also tested it with newest 8.4.2)

I use format/2 to print out a float number using ~ne format specifier. There I see some strange behavior:

format('~1e', [1e-3]).

This results in 1.0e-03, i.e., the exponent is zero-padded to two digits, which conforms to the C standard for printf function. The same holds for all n > 0.

However, when I call

format('~0e', [1e-3]).

I get 1e-003, i.e. the exponent has three digits now.

I know the print behavior depends on the used C library, but I find it someway inconsistent in the Windows version of Prolog.

I quickly checked the same with printf and some C compilers I have on my system: For gcc (7.4.0) it is always zero-padded to two digits and with Visual Studio (2010) it is always zero-padded to three digits.

So the question is: Is the reason for this inconsistent behavior the C library used in the pre-built Windows version or is there some special handling in the Prolog implementation?