floats default to NaN... why?

[F i L]

On Monday, 16 April 2012 at 04:05:35 UTC, Ary Manzana wrote:
> On 4/16/12 12:00 PM, F i L wrote:
>> On Monday, 16 April 2012 at 03:25:15 UTC, bearophile wrote:
>>> F i L:
>>>
>>>> I should be able to tackle something like adding a compiler 
>>>> flag to
>>>> default FP variables to zero. If I write the code, would 
>>>> anyone
>>>> object to having a flag for this?
>>>
>>> I strongly doubt Walter & Andrei will accept this in the main 
>>> DMD trunk.
>>
>> Do you have an idea as the reason? To specific/insignificant 
>> an issue to
>> justify a compiler flag? They don't like new contributors?
>>
>> I'll wait for a definite yes or no from one of them before I 
>> approach this.
>
> It's a flag that changes the behavior of the generated output. 
> That's a no no.

Don't *all* flags technically change behavior? -m64 for instance. 
How is this any different?

Besides, NaN as default is debugging feature. It's not the same 
thing as -debug/-release, but I think it makes sense to be able 
to disable it.