[rescue] NCD 88K Xterminal video adapter

[Mouse]

>> I can hardly claim to be representative of much, but *I* certainly
>> use [X's network-remoteness] routinely.
> Sadly, for most of the time Wayland has been around (or perhaps
> longer), the X community seems to have been madly destroying as much
> of the usability of remoteness as possible.

Not only of network-remoteness.  It was something like ten years ago I
ran into an application that not only assumed it was talking to a
TrueColor visual, but assumed it was 8/8/8 TrueColor, and, on top of
that, assumed it knew what the RGB masks were.  (Okay, it's possible it
would have errored out without TrueColor, or without 8/8/8 TrueColor.
But it _did_ misdisplay on a display that was RGB instead of BGR, or
whatever the orders the server provided and the client expected were.)

> Virtually everything other than xterm these days insists on rendering
> fonts server-side, making it utterly painful to use even over gig-E.
> Grumpf.

ITYPM "client-side".  Server-side is the normal (from an X perspective)
place to handle fonts.  But then, X font support cannot - and cannot be
made to - do antialiasing, flatscreen subpixel rendering, or the like;
it is specified too precisely, and specified in terms of displaying
1bpp glyphs.  You'd need extensions for such things.

As for being painful even over GigE, that's not just about client-side
fonts (though that probably doesn't help).  My terminal emulator will,
when told to, use a client-side font, and it is still fully usable
network-remote (visibly slower, but not by much).  But it pushes the
glyphs to server-side pixmaps, rendering them by copying from the
pixmap to the window.  I suspect what you're dealing with is rendering
all the way to a final image client-side and then pushing that, which
_is_ ridiculously inefficient except when (a) local and (b) using
MIT-SHM or moral equivalent (and also pushes respnsibility for handling
different visuals off onto the client).  The X.org world appears to
shoehorn everything into a Linux/x86 mindset, even to the point of
doing userspace bus enumeration for SBus SPARCs.  It's one reason I
still use MIT X when on hardware where it works, like my routine
screen-and-keyboard machines, which are SS20s.

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		mouse at rodents-montreal.org
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B