Re: Printing

[David De La Harpe Golden <david@harpegolden.net>]

> If that were so then only postscript experts would be able to teach
> their application to print something.
> 

Application developers use toolkits, toolkit developers target the 
printing system.  /Except/ on windows, that printing subsystem tends to 
involve postscript:  A toolkit, like gtk+ with cairo, draws vectors to a 
postscript document [1] instead of rasterising (or sending trapezoids 
(xrandr) or general polys (opengl) to the display driver in the modern 
era) to the screen when it wants to print, then sends that postscript 
document to the printing daemon which does whatever is necessary to 
print it (for cheap printers, maybe rasterising it fully to 
device-specific bitmap data with ghostscript, for postscript printers, 
maybe munging it a bit to the postscript level the printer wants and 
passing it on).  i.e. a hypothetical "pure gtk+ stack" emacs at the 
application level  might well use gtk+ and cairo for both screen display 
and printing and abstract over the individual platform and indeed screen 
vs. print differences, but cairo is still probably just going to be 
drawing to postscript underneath for printing purposes on free or even 
nonfree but unixy platforms.

Since emacs is currently a "toolkit unto itself" for the most part, as 
Stefan has already pointed out, it kind of makes sense for it to 
generate its own postscript.  Though maybe just saving out and 
"enscript"ing text files might currently work about as well...

On windows, the toolkit uses win32 drawing commands on a printer device 
context [2] instead of postscript drawing on a postscript document. 
Typically they might internally rasterise to a high-res DIB (.bmp) then 
printed with basically just one such win32 drawing command consisting of 
"draw this enormous .bmp please".

So, if you squint *a lot*, .bmp in the windows printing world is a 
*little* like .ps in the nonwindows printing world.  Maybe windows 
people might be conceptualising .ps as device-specific thing that things 
are printed _to_ (and it is that _too_ I guess), but it is /also/ a 
quasistandard intermediate format for printing in the relevant OS apis 
on nonwindows.

Mind you, PDF would work just as well or better than postscript for that 
purpose, but the current de-facto standard is for OSes to expose an 
interface that accepts postscript documents for printing. Windows is the 
only vaguely relevant one that _doesn't_, pretty much, and it's only a 
ghostscript installation away anyway, it provides mswinpr2 to rasterise
postscript to .bmps sent to native windows printers (which may then of 
course inefficiently convert them back to postscript to send  to native 
postscript printers, but anyway...)



[1] http://cairographics.org/manual/cairo-ps-surface.html
[2] http://cairographics.org/manual/cairo-win32-surface.html