Weird bug when loading spritesheets' sub-images

[nightm4re94]

^{This post has been migrated automatically from the old LITIENGINE forum (2018 - 2022).
Posted: 2019-06-29 20:36:54
User: CalvinMT [] (age: 1451 days 🔥; posts: 114)}
Hi!
I not sure if this bug is coming from my code or the LITIengine, that's why I'm posting it in this forum.
The bug happens rarely and randomly, and I couldn't find any way to reproduce it efficiently. Let me set an example.
I've got three files named ~~frame\_<name>\_unchecked.png and three other files name ~~frame\_<name>\_checked.png. They all look the same, except colorwise.
Each of them have ten sub-images with numbers to reprent levels from 0 to 9.
Sometimes, or more rarely and randomly as said previously, in-game, one sub-image won't have the right color nor the right number.
Here's the only part of the code that deals with those sub-images:
`[code] private void addLevels () {
this.nbOfLevelsInSelectedWorld = 0;
Resources.maps().getAll().stream().forEach(map -> {
if (map.getName().startsWith(this.worldSelect.getName().toLowerCase()) && this.nbOfLevelsInSelectedWorld < this.MAX_LEVELS_PER_WORLD) {
this.nbOfLevelsInSelectedWorld++;
}
});
this.listLevelsShowing.stream().forEach(level -> {
level.setVisible(false);
});
this.listLevelsHoveredImage.stream().forEach(hoveredImage -> {
hoveredImage.setVisible(false);
});
Image frameImageHovered = (Image) Resources.spritesheets().get("frame_" + this.worldSelect.getName().toLowerCase() + "hovered").getImage();
for (int i=0; i<this.nbOfLevelsInSelectedWorld; i++) {
Image frameImage;
switch (this.worldsData.get(this.worldSelect.getName()).getListLevelState().get(i)) {
case UNCHECKED:
frameImage = (Image) Resources.spritesheets().get("frame" + this.worldSelect.getName().toLowerCase() + "unchecked").getSprite(i);
break;
case CHECKED:
frameImage = (Image) Resources.spritesheets().get("frame" + this.worldSelect.getName().toLowerCase() + "_checked").getSprite(i);
break;
default:
frameImage = null;
break;
}
this.listLevelsShowing.get(i).setImage(frameImage);
this.listLevelsHoveredImage.get(i).setImage(frameImageHovered);
this.listLevelsShowing.get(i).setImageScaleMode(ImageScaleMode.FIT);
this.listLevelsHoveredImage.get(i).setImageScaleMode(ImageScaleMode.FIT);
this.listLevelsShowing.get(i).setImageAlign(Align.CENTER);
this.listLevelsHoveredImage.get(i).setImageAlign(Align.CENTER);

this.listLevelsShowing.get(i).setVisible(true);
}
}
[/code]`
I've checked everything with prints. Even ~~*~~Spritesheets.get()* and ~~*~~Spritesheet.getSprite()*. And everything's correct.
So my last conclusion is: their could be, sometimes, that a pointer points towards two different sub-images. And I was wondering if [b]~~*~~parallelStream()*[/b] would be responsible, or something somewhere in [b]~~*~~Resources.load()*[/b]...?

[nightm4re94]

^{This post has been migrated automatically from the old LITIENGINE forum (2018 - 2022).
Posted: 2019-07-03 15:49:47
User: CalvinMT [] (age: 1451 days 🔥; posts: 114)}
I know there's no pointer manipulation on the user side. But as the problem looked as if it was a pointer problem, I had to check the ~~*~~Resources.load()* method in case that was the problem. But, as anyone could guess, it wasn't. Although it doesn't seem like that's where sprites and spritesheets are really loaded...
I'd like to know where [b]~~*~~sprites*[/b] and [b]~~*~~spritesheets*[/b] get loaded from the [b]~~*~~litidata*[/b] file so I can do some more checks. I couldn't find it in ~~*~~Resources.load()* and I'm having trouble finding it elsewhere.
So where can I find it?

[nightm4re94]

^{This post has been migrated automatically from the old LITIENGINE forum (2018 - 2022).
Posted: 2019-08-03 12:46:12
User: CalvinMT [] (age: 1451 days 🔥; posts: 114)}
The image starts being the wrong image in the method ~~getImage() from **~~ImageComponent.java**. At this line to be precise:
~~```~~Optional<BufferedImage> opt = Resources.images().tryGet(cacheKey);```` After that line, I checked ~~~~opt's colors and it returned colors that weren't supposed to be on the image to be shown. Instead, other colors from an other image that I've got were returned. And that image was the (wrong) image that was displayed. I've verified ~~~~cacheKey and it's fine. So my guess is that ~~~~tryGet()` returns the wrong image for some reason.

[nightm4re94]

^{This post has been migrated automatically from the old LITIENGINE forum (2018 - 2022).
Posted: 2019-08-03 15:51:14
User: CalvinMT [] (age: 1451 days 🔥; posts: 114)}
I've been digging some more and managed to make my way up to ~~public T get(URL resourceName, boolean forceLoad) from **~~ResourcesContainer.java**.
Before the method returned its value, I asked it to print out the colors of what's in ~~~~resources with the string key given as ~~~~resourceName as followed:
~~```~~ public T get(URL resourceName, boolean forceLoad) { // [...] } else { // goes through else because 'forceLoad' is false if (resources.containsKey(resourceName)) { System.out.println(new Color(((BufferedImage) resources.get(resourceName)).getRGB(0, 0)).toString()); } return this.resources.computeIfAbsent(resourceName, this::loadResource); } }```` As expected, the print shows the wrong colors. So it appears that either ~~~~resources doesn't give the right image, but I doubt it. Or it is given the wrong image while loading resources. That or ~~~~resourceName` is wrong and I'm going down the wrong track.

[nightm4re94]

^{This post has been migrated automatically from the old LITIENGINE forum (2018 - 2022).
Posted: 2019-08-06 23:17:28
User: CalvinMT [] (age: 1451 days 🔥; posts: 114)}
I'm having trouble pinpointing that bug. I decided to make a simple program which can reproduce it so you guys can have a look: https://github.com/CalvinMT/LitiengineSpritesheetBug
For information, I noticed that the bug wasn't there in release v0.4.17-alpha of LITIengine. But appeared during the development after that release (i.e.: during the developement of v0.4.18-alpha). So you'll have to add the currently "work-in-progress" version of the LITIengine to the program's build path.
And don't forget to ~~__read the **~~README.md__** file!

[nightm4re94]

^{This post has been migrated automatically from the old LITIENGINE forum (2018 - 2022).
Posted: 2019-08-10 18:25:34
User: nightm4re [Admin 👑, Mod 🔧] (age: 1567 days 🔥; posts: 185)}
Thanks for making this reproducable, we'll hopefully figure it out this weekend. Cheers!