Re: embedding GTK<->Qt: they can, can we ?

[Federico Mena Quintero <federico helixcode com>]

"Billy N. Patton" <bpatton dal asp ti com> writes:

> My needs are to display/future edit integrated circuits.  Currently
> I'm dealing only with cells levels 100-> 20,000 transistors.
> 
> Withing the next 2 years my programs (QC) will need to handle the
> ascii files that eill be > 4gig.  I told this to the
> wx list and got one response that said that wx was not meant for
> that much data.  Their new (under development) canvas widget would be
> great for a small amout of data.  But you must create a rectangle,
> polygon, circle and others with their classes for their canvas
> to manage it.  2 copies of a 4 gig file in memory, I DON"T THINK SO!
> 
> gtk provides no internal canvas database.  I ptovides no drastically
> shorthand methods for controlling the screen or data.  It is just
> a generic call to the X functions.  Much like programming in Xt.
> 
> COuld you imagine the amount of memory necessary to display 4 million
> rectangles if I had to maintain my data for output and a canvas maintained
> its data for display :(
> 
> What I really need is what the old Apollo's had.  A set of
> functions that directly accessed the graphics and did not go through
> an X server.
> 
> I do not intend to send 4M draw_rectangles to the X server.  I draw
> to pixmap the send pixmap to server.

If you need to draw that much data, do it in the client and draw the
stuff to an RGB buffer or similar.  Then draw that to the X server
using the GdkRGB calls.

The Guppi plotting framework does pretty much just that, and it can
plot huge amounts of data very quickly and efficiently.  You can look
at it to see how it does things.

  Federico