Yes, Float is a bad name for Float64 since anyone coming from C or Fortran 
would expect Float = Float32 and Double = Float64.

If you're writing Float64 a lot I think you may be over constraining your type 
signatures. Most algorithms that make sense for Float64 also make sense for 
Float32 and BigFloat and possibly also for integers and rationals, maybe all 
real number representations. Unless you very specifically need a 64-bit float 
for your code to work, why restrict it more than necessary? And if you really 
need a 64-bit float – not a 32-bit one or a 256-bit one – then it seems quite 
fitting to have "64" in the type signature.

> On Jul 29, 2014, at 6:50 PM, John Myles White <[email protected]> 
> wrote:
> 
> One of the things I like about the Julia community is a broad preference for 
> clarity over brevity. Think of it as the opposite of the Perl culture.
> 
> In this case, Float would be less, rather than more, clear because our Float 
> would describe a type that most languages would call Double.
> 
>  -- John
> 
>> On Jul 29, 2014, at 3:45 PM, Júlio Hoffimann <[email protected]> 
>> wrote:
>> 
>> 2014-07-29 19:32 GMT-03:00 John Myles White <[email protected]>:
>>> I think the confusion is that Julio assumes Int is used for brevity, when 
>>> it is actually used for cross-platform compability.
>>> 
>>>  -- John
>> 
>> Yes, I assumed the Int alias had these two goals: brevity + portability. I 
>> still think a cleaner Float alias would be useful to avoid typing 64 
>> everywhere.
>> 
>> Júlio.
> 

Reply via email to