Printing floating points

[Bruce Carneal]

On Tuesday, 26 January 2021 at 20:34:51 UTC, Ola Fosheim Grøstad 
wrote:
> On Tuesday, 26 January 2021 at 20:27:25 UTC, Ola Fosheim 
> Grøstad wrote:
>> For instance, for a converter from int to string, the most 
>> likely failure point will be tied to crossing between "9" and 
>> "0" as well as the ends of the base-2 input. Meaning, if it 
>> works for 1,2,3, you know that it also works for 4,5,6,7,8, 
>> but maybe not 9,10.
>
> Or to put it another way. If you can prove that these 
> implication holds:
>
> 0-1 works => 2-8 works
> 0-11 works => 12-98 works
> 0-101 works => 102-998 works
> etc
> then you only need to test 9-11, 99-101.

I think this is a great way to go for simple functions.  If the 
domain ==> co-domain mapping is truly obvious then we can exploit 
that to write an efficient unittest that we can take as the next 
best thing to the simpler exhaustive proof.

I don't see a mapping for the current problem that would allow 
for such efficient enumerative testing but it's not my field.  
Still, if/when you prove that you've fully covered the domain I'd 
be very excited to read about it.  It could be a major advance.  
Note: by "proof" I do not mean probabilistic evidence.  We 
already have that.

Even if you don't come up with proof, the attempt may be worth 
your time.  Good luck.