| Backing Store Question [message #45465] |
Fri, 16 September 2005 06:35  |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Folks,
In poking around in the IDL documentation this morning, I came
upon this in the WIDGET_DRAW documentation:
On some systems, when backing store is provided by the
window system (RETAIN=1), reading data from a window
using TVRD( ) may cause unexpected results. For example,
data may be improperly read from the window even when
the image displayed on screen is correct. Having IDL
provide the backing store (RETAIN=2) ensures that the
window contents will be read properly.
Anybody have any idea which systems this is taking about?
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
|
|
|
|
| Re: Backing Store Question [message #45645 is a reply to message #45465] |
Fri, 16 September 2005 13:09  |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Karl Schultz writes:
> The server may be able to repair with backing store at one
> point in time, and then may fail to do so later if resources (memory) get
> too tight. So, the entire backing store functionality is a little bit
> unsteady in this respect. I think that backing store may have been more
> popular and important in older systems when it took longer to render
> things. Nowadays, I'm a little hard-pressed to find X servers that can be
> configured to use backing store. I know that some of the latest Xorg
> servers do not support it.
I'm really surprised by the last two sentences. I would have
thought that with computer memory so cheap that this would
be a common service of any X server. I'm surprised to hear
there are modern X servers that don't support it at all.
So, bottom line, are direct graphics so "1970s" that they
won't even run on modern computers? No wonder RSI is spending
all my maintenance dollars on object graphics! I'm beginning
to see the light. :-)
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
|
|
|
|
| Re: Backing Store Question [message #45647 is a reply to message #45465] |
Fri, 16 September 2005 12:25 |
Karl Schultz
Messages: 341
Registered: October 1999
|
Senior Member |
|
|
On Fri, 16 Sep 2005 07:35:53 -0600, David Fanning wrote:
> Folks,
>
> In poking around in the IDL documentation this morning, I came
> upon this in the WIDGET_DRAW documentation:
>
> On some systems, when backing store is provided by the
> window system (RETAIN=1), reading data from a window
> using TVRD( ) may cause unexpected results. For example,
> data may be improperly read from the window even when
> the image displayed on screen is correct. Having IDL
> provide the backing store (RETAIN=2) ensures that the
> window contents will be read properly.
>
> Anybody have any idea which systems this is taking about?
>
It is referring to X servers. On Windows, setting RETAIN to 1 effectively
sets RETAIN to 0.
The intent behind RETAIN=1 in the X environment is to use the X server's
backing store to repair the window when it needs repairing. The idea is
that when a window is damaged, the X server tries to repair it itself from
bits it may have stored in backing store. If the server can't do it, it
will send an Expose event to the application, which means running the IDL
application's expose event hander, if there is one. If you didn't code an
expose handler, hoping that RETAIN=1 would always repair the window for
you, then you'll be out of luck if the X backing store mechanism fails for
some reason.
The server may be able to repair with backing store at one
point in time, and then may fail to do so later if resources (memory) get
too tight. So, the entire backing store functionality is a little bit
unsteady in this respect. I think that backing store may have been more
popular and important in older systems when it took longer to render
things. Nowadays, I'm a little hard-pressed to find X servers that can be
configured to use backing store. I know that some of the latest Xorg
servers do not support it.
I suppose that the warning about the TVRD is pretty much the same issue as
being able to repair the window. The server could respond to a read
request with pixels from the backing store, but if it had to give up on
providing the backing store, it wouldn't be able to give back those
pixels. I'm a little confused about why reading them off an undamaged
window wouldn't work, but I think that's the general idea.
Karl
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
