Supercollider dev question

[Marc Nostromo [M-.-n]]

Template specialisation. I never know really how to write them.

For example, I'm writing the template class for fixed point that works like

typedef FixedFP<int32,28> sample_t ;

where the first argument is the 'container type' (how we're going to store
the fixed point) and the second is the number of bits used for the
fractional part

When you do a multiply, you need a 'temporary' container that is twice the
number of bits from the original container so that if it is int32 it uses
int64 if it is int8 uses int16 and so on.

There are several ways of doing: on is to re-implement the operator* for
each supported type with the typename explicitely stated or to have a
mechanism that for one type, gives you the type of the 'expanded containter'
type. Both need template specialisation which is fine in theory but, since
i'm not 100% template HxC, I can't type it right the first time and need
countless iteration before the compiler accepts it :(

At least, now that's sorted, I got my basic class working. I'll need to
polish it a bit and add support for operations between FP with variable
fractional parts but it's not really though. Then in a few years, I can get
Tony to fiddle with multiply dithering.

Now to change every reference of sample_t in my framework and pray it still
works :)

/M
PS: There's a (overly) extensive description of a FP class here
http://www.codeproject.com/KB/cpp/fp_math.aspx for any one that is willing
to scratch it's head very deeply


"Thanks" to you I'm going for a template based mechanism.
>>
>> Boy I HATE c++ template syntax :)
>>
>
>
> - Peter
>
>
> _______________________________________________
> music-bar mailing list
> music-bar at lists.music-bar.org
> http://lists.music-bar.org/cgi-bin/mailman/listinfo/music-bar
>
>


-- 
http://nostromo.noisepages.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.music-bar.org/pipermail/music-bar/attachments/20111027/82634b86/attachment-0002.html>