right, or just loop through the existing types and check whether 8*sizeof() 
is large enough.

i didn't find anything in base either.  the reason why o thought it should 
perhaps be added is because it requires knowledge of hat sizes are 
supported.

andrew

On Monday, 30 December 2013 12:24:20 UTC-3, John Myles White wrote:
>
> I’m still a little confused, but it sounds like Elliot’s suggestion was 
> right: iceil(log2(x)) will tell you how much bits are needed to represent 
> the value, so you’d need some function that wraps that calculation and then 
> finds the closest power of 2 to the output to determine the right storage 
> type. 
>
> I’m pretty sure that function doesn’t exist in Base. 
>
>  — John 
>
> On Dec 30, 2013, at 6:35 AM, andrew cooke <[email protected] <javascript:>> 
> wrote: 
>
> > 
> >     julia> smallest_uint(5) 
> >     Uint8 
> > 
> > so this isn't typeof() (which you might have been describing). 
> > 
> > i have code that needs to generate a table of values (bit fields, but 
> Vector{Bit} isn't so useful), but the number of the bits can vary. 
> > 
> > andrew 
> > 
> > On Monday, 30 December 2013 00:00:11 UTC-3, John Myles White wrote: 
> > I’m a little unsure what you mean. Are you asking for a function that 
> given a 32bit value returns that it’s of type Uint32? I would guess (but 
> maybe am way offbase) that you’ll be storing 30bit unsigned integers inside 
> of Uint32’s, right? 
> > 
> >  — John 
> > 
> > On Dec 29, 2013, at 7:45 PM, andrew cooke <[email protected]> wrote: 
> > 
> > > question says it all really.  if i need to deal with 30bit unsigned 
> integers, is there anything that returns Uint32? 
> > > 
> > > [obviously i can just tabulate what there is if missing, but i 
> wondered if it would also be good for the std lib.] 
> > > 
> > > thanks, andrew 
> > 
>
>

Reply via email to