On Wed, Jun 18, 2008 at 2:04 PM, Andrew Lentvorski <[EMAIL PROTECTED]> wrote:
> SJS wrote:
>>
>> Um... the postscript patent has expired. At least postscript level 1 can
>> be implemented without encumbrances. Level 2 should be expiring soon.
>>
>> You'd think that the expensive of adding a postscript interpreter to
>> a printer would be minimal in this day and age.
>
> The problem is RAM. Printers are being ever pushed to be cheaper while, at
> the same time, being pushed to handle larger and larger documents in terms
> of resolution.
>
> A Postscript printer has to keep, at minimum, 8.5x11x600x600 bytes of RAM,
> about 33 megabytes.
You just allowed one byte per pixel. But in fact laser printers work
at one _bit_ per pixel, they don't print different intensities of
those little spots. They fake greyscale by halftoning, changing the
density of filled-in spots.
Printer manufacturers have algorithms that store the rendered
PostScript image in compressed raster form, and expand the raster data
on the way to the print engine.
> More if we have more that 8-bit grayscale or want to do
> color conversion post-rendering. 33 Megabytes doesn't fit on a nice cheap
> System on a Chip. So, instead of having a single chip to control
> everything, you need a chip with a RAM interfaces and a stick of RAM
> (probably cheaper than the chips).
>
> Thus, we wind up with printers that try to do as little as possible and
> force the host system to do the rendering.
>
> To be fair: I don't find that a bad tradeoff if the manufacturers would just
> document the raster protocol. A printer is almost always going to have less
> power and memory than a computer.
Early counterexample -- the Apple Laserwriter had much more processing
power and memory than the computers it was originally connected to.
Early (undocumented) example. The NeXT laser printer was a pure print
engine, with rasterized data sent to it over a proprietary balanced
high-speed serial line. The computational effort to render PostScript
and generate the data did not noticeably slow down a 25 MHz 68030.
One interesting undocumented feature was that the sync signal from the
printer to the computer traveled over the same cable, so a cable
extender had to be a half-wavelength so the data and sync would slip
by a whole signal period.
> The issue is that printer manufacturers won't document their raster
> protocols, and Postscript gains a benefit from being a lingua franca.
carl
--
carl lowenstein marine physical lab u.c. san diego
[EMAIL PROTECTED]
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
