tracking down high memory usage in skiptool

[pastro]

The WizUiList was basically cloned from sub's UiList object, and uses the same caching scheme that he uses. If the cache fills, it reuses the oldest element and sets forceRefresh to true.

The bitmaps are returned in a wizuilabel not a wizuilist. When I want to change the image I use something like


lblProgramInfo.addArg("@thumbnail", targetProgram);

where lblProgramInfo was defined the object as
privateWizUiLabel lblProgramInfo;

And targetprogram was defined as an object of showinfo:
publicclassShowInfo : SkinHelper2.GetImageCallback


I know this is pretty hard to follow but if this means anything to you guys let me know.

[JavaWiz]

The bitmaps are returned in a wizuilabel not a wizuilist. When I want to change the image I use something like


lblProgramInfo.addArg("@thumbnail", targetProgram);

where lblProgramInfo was defined the object as
privateWizUiLabel lblProgramInfo;

And targetprogram was defined as an object of showinfo:
publicclassShowInfo : SkinHelper2.GetImageCallback


I know this is pretty hard to follow but if this means anything to you guys let me know.

I think I may have found a bug in WizUiLabel and WizUiTextBox that may attribute to this memory problem.

I've attached a fix version of the DLL, but ask that you don't deploy it to the public until I get more testing done as I think there still may be some tweaking to be done.

Let me know if this helps with your memory problems...

[pastro]

With the updated DLL, I get a lot of getimage callbacks but the bitmaps never change on the screen. It always displays the first image and nothing else.

[pastro]

Actually let me change that comment. In skiptool where I don't set forcerefresh to true, I don't get any new bitmaps displayed. In burndvdx2 where I experimented and always set forcerefresh=true I do get the updated bitmaps to display, and the memory does not climb either. There is a solution in here somewhere.

I know you are going to puke when you see this but here is the code that goes through and sets forcerefresh

The check for null was something I had to do because occasionally I get a null image from wizutilities when I am enabling and disabling buttons. That's a different problem though.


for (i = 0; i < renderList.Count; i++)
{
testme1 = (GBPVRUiElement)renderList[i];

if (testme1.image != null)
{
if (testme1.name == "ProgramInfo")
testme1.forceRefresh = true;
arraylistme.Add(testme1);
}

[JavaWiz]

With the updated DLL, I get a lot of getimage callbacks but the bitmaps never change on the screen. It always displays the first image and nothing else.

Try again...

I've actually added code similar to that right in the WizUi controls, so you shouldn't have to.

[JavaWiz]

The check for null was something I had to do because occasionally I get a null image from wizutilities when I am enabling and disabling buttons. That's a different problem though.


for (i = 0; i < renderList.Count; i++)
{
testme1 = (GBPVRUiElement)renderList[i];

if (testme1.image != null)
{
if (testme1.name == "ProgramInfo")
testme1.forceRefresh = true;
arraylistme.Add(testme1);
}

Is this code in the GetRenderList of your plugin?

I caught a similar exception with some buttons, so the button code now checks for a null before adding to the list. You may be able to remove this code with the next version of WizUiHelper.

[pastro]

Is this code in the GetRenderList of your plugin?

I caught a similar exception with some buttons, so the button code now checks for a null before adding to the list. You may be able to remove this code with the next version of WizUiHelper.

yes that is in getrenderlist. It was sort of a long standing bug that this fixed. I'll check if I get any issues if I remove it.

[pastro]

That fixed it. The memory stays constant and doesn't climb and the bitmaps all show up without my forcerefresh=true.


It will take longer to determine if the null problem still exists, but I will test that further and post the results.
Nice work. The UI is much more responsive after I use the control for a bit.

[sub]

Great - well done.

I'm glad to see I'm off the hook too, with no major memory leak in the framework.

[pastro]

Null problem appears fixed as well.
Thanks a bunch.
Give me the high sign when it is ok to release.