[PD] Question about arrays on 64-bit

=?X-UNKNOWN?Q?g=FCnter_geiger?= geiger at xdv.org
Sun Mar 20 22:24:36 CET 2005


On Sat, 19 Mar 2005, thewade wrote:
> Hello list!
>
> I was wondering if someone could explain why arrays in PD on 64-bit
> machines are messed up? I see in the bug on sourceforge that there
> are a couple of sub-optimal ways to solve this: what does that mean?
>
> Shouldnt it be as easy as changing the array pointers from int to
> long? (I havent looked at the code and I think that I would just
> get lost if I did. Im not very good at programming.)

No, its not that easy. Maybe I find some time to implement a solution
and upload a diff this week, so you can at least have the "hacky"
solution.

> Thanks for the explination in advance!

Here it goes:

Array don't store floats but unions of t_word, which are 64 bit broad
on 64 bit architectures.
- One solution is to store doubles instead of float and change all the
types of the temporary variables in all objects accessing array to
double. (externals using float arrays have to be adapted too)
- Another solution is to ignore the t_word type and store all values
as floats in arrays (and read them as floats like it is now)
- We could implement a special iterator or access function to float
arrays (instead of ++ or []) and change all the occurrencers of all
float array objects according to this. (Breaks all float array externals).

The second solution is the ugliest one, but would keep all float array
reading objects happy. I actually had that implemented when I was
regularily using the 64 bit alpha AFAICR.

Guenter

(PS: written down from memory. I would be happy to be wrong and there
are better solutions)


> -thewade
>
> _______________________________________________
> PD-list at iem.at mailing list
> UNSUBSCRIBE and account-management -> http://iem.at/cgi-bin/mailman/listinfo/pd-list
>





More information about the Pd-list mailing list