Resolution vs hinting [was: OpenGL, xserver, and anti-aliased fonts]

If you'll excuse the segue, this relates to a question I was
going to post someplace tonight....

At what point does it no longer make sense to bother with hints
and to use just the outline as a vector when rendering text?

Is 600 dpi at 24 bits per pixel sRGB enough?  300 dpi at 48?

Should the display hardware make an effort to snap to pixels
irregardless of the displays resolution and depth?  And if
so should it happen at the panel rather than at the cpu?

200 dpi panels appear to be on the market (I beleive sharp has
a 640x480 panel the same size as eg the ipaq's 320x240 panel,
roughly 3.2 by 2.4 inches).  The 1920x1200 panels are closing
in on 200 dpi (I've seen 180 dpi mentioned).  300 should be
readily doable.  I can't imagine that 600 will be all that
far behind; perhaps only a few years.

Also, is it reasonable to use modern ink jets to model this,
given that most drivers prefer to send the data as sRGB and
let the printer convert to CMYK (or CcMmYK, CcMmYKk, etc).

At some px/em the fonts tend to stop deforming the outlines,
yes?  Do full colour or grayscale displays provide enough
extra signal to ignore the deformations before the fonts do?

-JimC