Re: preferences->preview->thumbnail size

[Joshua Judson Rosen <rozzin geekspace com>]

Alexander Larsson <alexl redhat com> writes:
>
> On Thu, 2008-12-11 at 08:14 -0500, Joshua Judson Rosen wrote:
> > "Dr. Michael J. Chudobiak" <mjc avtechpulse com> writes:
> > >
> > > Richard Hughes wrote:
> > > > cc1: warnings being treated as errors
> > > > nautilus-global-preferences.c:99: error: this decimal constant is unsigned only in ISO C90
> > > > nautilus-global-preferences.c:100: error: this decimal constant is unsigned only in ISO C90
[...]
> > Alternately, it could be a logarithmic scale, which would be
> > fine-grained in the kilobyte range and courser in the gigabyte range
> > (because 1 kB +/- 100 bytes is significant; 1 GB +/- 100 bytes, on the
> > other hand...).
>
> I don't disagree, but that would mean changing the way a deployed
> preference setting works, so it would mean either breaking old settings
> or doing some contortions to handle both. I don't think its worth the
> pain for either really.

What about just using floats instead of uints? It'd still act like a
linear scale, even though IEEE floats are actually stored (in the
hardware) logarithmically; but you'd be able to store these
huge, low-precision integers and also the high-precision small
integers.

I know it's a little strange to be suggesting floats for their
*internal scale* characteristics rather than their ability to
represent fractions (with no fractions in sight, even!), but....

Is it too much pain to move from using uints to using floats?

-- 
Don't be afraid to ask (Lf.((Lx.xx) (Lr.f(rr)))).