[Xcb] XCBImage

Jamey Sharp jamey at minilop.net
Mon Mar 14 23:31:21 PST 2005

On Tue, Mar 15, 2005 at 07:43:09AM +0100, Vincent Torri wrote:
> anyway, i prefer separate declaration and initialization. I know that i've
> read something about that, but i can't remember what it is.

Heh. Well, when your account gets created, I guess you can change it if
you want to. :-)

Personally, I usually prefer to use initializers whenever C allows me
to, unless doing so makes the code hard to read.

> for the test programs, i'm writing a small app that should load a png
> file, display it in a window and create a new small window that displays a
> zoom of a part of the image around the mouse pointer. That should be
> enough to show how to use (put/get)image and (put/get)pixel. I'll try to
> use shm also.

Sounds like a cool demo. I'll take it. :-)

> On Mon, 14 Mar 2005, Jamey Sharp wrote:
> > I notice you also changed GetPixel and PutPixel to take INT16s instead
> > of ints for the x and y coordinates. I think that's probably a bad idea.
> > People might want to deal with images larger than 64k pixels wide or
> > tall. So I'm not committing it yet either, but I'll happily listen to
> > arguments.
> yes, you're right. Using INT32 is enough ? (i try to use only X defined
> type at most as posible)

Yeah... when people can afford to throw 16GB of RAM at a single row of
their images, we can think about changing the API.

Probably it should be unsigned (a CARD32) though. Allowing negative x or
y coordinates makes sense in general, but I think can only produce
surprising results here.

I'm curious: why do you prefer X types? I don't have a problem with that
preference, but I don't understand it.


More information about the xcb mailing list