Re: "A diagram of C23 basic types"

On 18/04/2025 03:39, Lawrence D'Oliveiro wrote:

On Mon, 7 Apr 2025 22:46:49 +0100, bart wrote:
 

(In source code, it would also be useful to use 1e9 or 1e12,
unfortunately those normally yield floating point values.

 Tried Python:
      >>> type(1e9)
     <class 'float'>
     >>> round(1e9)
     1000000000
     >>> round(1e12)
     1000000000000
 However:
      >>> round(1e24)
     999999999999999983222784
 So I tried:
      >>> import decimal
     >>> decimal.Decimal("1e24")
     Decimal('1E+24')
     >>> int(decimal.Decimal("1e24"))
     1000000000000000000000000
 which is more like it.

The idea behind writing 1e12 for example was for something that was compact, quick to type, and easy to grasp. This:
    int(decimal.Decimal("1e24"))
seems to lack all of those. Besides:
   import decimal
   def fn(n):
       for i in range(100_000_000):
           a = int(decimal.Decimal(n))
       print(a)
   def fn2(n):
       for i in range(100_000_000):
           a = round(n)
       print(a)
   def fn3(n):
       for i in range(100_000_000):
         a = n
     print(a)
   fn("1e24")       # 50   seconds (loop overheads excluded)
   fn2(1e15)        # 13   seconds
   fn3(1e15)        #  0.6 seconds
You don't want it to be much slower either: it should not affect performance. (Timings for CPython.)