[DGD] Full double precision floats?

[Raymond Jennings]

Since we're dealing with floating point formats that have the same exponent
range, I presume that silent interconversion (with an appropriate warning
about mantissa truncation on going back to 6-byte floats) will suffice.

My hope was to convert dgd to use full double precision floats on a
permanent basis.  Besides increased precision, I figured it would simplify
the floating point code.


On Sat, Apr 27, 2013 at 3:53 PM, Felix A. Croes <felix at dworkin.nl> wrote:

> Raymond Jennings <shentino at gmail.com> wrote:
>
> > Making it optional will complicate things.  Is there room in the snapshot
> > format for a flag instead of just incrementing the dump version?
>
> You shouldn't increment the dump version, you can reserve a bit in
> the dflags field (0x08 will do).
>
> If you don't want to be compatible with the mainline, you can work
> on a separate branch.
>
> Regards,
> Felix Croes
> ____________________________________________
> https://mail.dworkin.nl/mailman/listinfo/dgd
>