[Pdl-porters] BAD values and longlongs

[Craig DeForest]

With all the discussion of the difficulty of storing extra-long ints in double precision floats, I got to thinking -- inside the PDL structure itself, the badval is stored as a double -- always. Do we need to update the badval infrastructure so that the badval is stored as the same type as the PDL?

[Chris Marshall]

Yes, if you'll look at the bug ticket for longlong truncation on sf.net,
I've added comments to that effect. As mentioned in the bump to
perl 5.10.x thread, I would like to close out the development with
support for perl 5.8.x with a fully working PDL-2.x with 64bit index
support.

--Chris


On Mon, Nov 25, 2013 at 10:00 AM, Craig DeForest

Post by Craig DeForest
With all the discussion of the difficulty of storing extra-long ints in double precision floats, I got to thinking -- inside the PDL structure itself, the badval is stored as a double -- always. Do we need to update the badval infrastructure so that the badval is stored as the same type as the PDL?
_______________________________________________
PDL-porters mailing list
http://mailman.jach.hawaii.edu/mailman/listinfo/pdl-porters