How do I avoid performance-robbing operations while resizing a form?

  • Thread starter Thread starter Peter Duniho
  • Start date Start date
P

Peter Duniho

The basic idea: I've got a form in which I draw a bitmap, scaled to fit
the form. For obvious reasons, I want to cache the scaled version of the
bitmap so that repainting the form is faster. I need to update the cache
any time the form size is changed. But I don't want to generate the
scaled bitmap every time I get the Resize event, because doing so makes
the resizing itself way too slow, even on a fast computer.

Ideally, I'm looking for an event that basically says "the form size has
changed and the user is done dragging". Alternatively would be some
convenient way to detect when the user starts and stops dragging the form.

The only solutions I've come up with so far are getting into the nuts and
bolts of the form's window, with the WndProc and the window messages
(probably have to be non-client, since I don't see anything obvious in the
basic WM messages that would work), or taking a completely different
approach to maintaining performance of the Resize event.

I'd prefer to avoid messing around with the WndProc, especially if there's
a better .NET Framework-approved way of dealing with the issue. In the
latter idea, all I can come up with is doing the bitmap caching in the
background on a different thread. That would work, and probably pretty
well too, but it seems overly complicated.

It seems like this would come up fairly often...what techniques are other
people using to deal with the problem?

Thanks!
Pete
 
[...]
In the latter idea, all I can come up with is doing the bitmap caching
in the background on a different thread. That would work, and probably
pretty well too, but it seems overly complicated.
Click to expand...

For the record, I tried this mechanism, and it didn't improve
responsiveness at all anyway. So, obviously my guess of "pretty well" was
way off. :)

I admit, even without rebuilding the cached bitmap, performance for
resizing the form isn't great. But the best responsiveness still comes
when I don't do anything except redraw (no creating new bitmaps). I'd
sure love to hear if anyone has ideas for how to either improve the
overall responsiveness, or know when the user is dragging the window (so I
can avoid bothering to rebuild the cached version when that's happening).

Thanks,
Pete
 
why not used vector graphic instead of a bitmap?
i.e. instead of generating a bitmap, why not draw directly on the form?

--
Regards,
Lloyd Dupont
NovaMind Software
Mind Mapping at its best
www.nova-mind.com
 
why not used vector graphic instead of a bitmap?
i.e. instead of generating a bitmap, why not draw directly on the form?
Click to expand...

Because the image that I'm drawing is itself a bitmap. I agree it would
be a waste to cache non-bitmap data to a bitmap, but since I'm starting
with a bitmap in the first place, it's not an option to just render using
vector-based commands.

Pete
 
Are you using .NET 2.0? Forms have ResizeBegin and ResizeEnd events.
Click to expand...

Yes. I missed those. Probably because I was looking only at the events
in my custom control, rather than in the form containing that control.
Thanks!

Pete
 
Yes. I missed those. Probably because I was looking only at the events
in my custom control, rather than in the form containing that control.
Thanks!
Click to expand...

Just to follow up...

These events do just what I want. By setting a flag in ResizeBegin, I can
conditionally do the extra work either in the OnClientSizeChanged method
(if the flag is clear) or in the ResizeEnd event handler. This way I only
do the expensive work of caching a new bitmap the right size for the
control when the user is done messing with the window.

One little gotcha: you only get these events in a Form, but I'm using them
in a custom Control. So my Control-derived class has to subscribe to the
ParentChanged event, walk up the tree to find the actual Form that owns
the control, and then subscribe to that form's ResizeBegin and ResizeEnd
events.

Seems a little clunky to me, but it works. :) Thanks for the tip Brian.

Pete
 
Back
Top