Yes, replacing my map entity with a model does not increase the nexus on each ent_create.

Dooley, I think you are chasing the wrong issue here. A memory leak doesn't lead to memory corruption. Leaking memory isn't good and should be fixed, but it's not the cause of your garbage data. That usually comes from writing into corrupt pointers and overriding existing data.

Possibly. However, the memory leak is causing the game to crash too. If I can fix it, at least I can move on to the next possible cause...

Oh man, you were right. I did a great job of reducing the amount of memory the game uses. Memory usage seems to stabilize much lower than the previous version of the game.

However, the data corruption is still happening.

Okay, I am opening up a file, reading from it, and storing the information in variables. Then, it re-writes the file with updated information and saves it again. So I guess this may be what you described as "overriding existing data."

Can you tell me if there is a right or wrong way to do this? If it's as simple as just not over-writing a file, I may be able to come up with a better way to do it.

Okay I found a potential cause of corruption.

For some reason, when I wanted to check if a file existed, I was using

if(file_open_read(filename))

instead of

if(file_exists(filename))

In most cases this actually worked okay, since the function was looking for an existing file. However, in one case, it would only work when the file did not exist.

In this case, if the file did exist, I guess that file would actually remain open for reading, and it never got closed. I am about to test the game with this corrected ... wish me luck!

Download process hacker or any other tool and check for open (file) handles.
You will spot 99% of all not closed handles instantly

And yes, your above code was leaking a handle on ANY valid file because you didnt save the handle/return value...

However, in theory, this IS the way to go to check if a file exists IF you want to read it as well!

The reason is that otherwise you might end up with a race condition: You check if the file exists and then open the file. Between the check and you opening the file, the file gets removed and you end up trying to open a file that doesn't exist (which returns an invalid handle).

The right way to do it is to open the file, which checks if it exists anyway, and then to check the returned file handle. And as Chaos pointed out, don't forget to close the files once you are done reading them, or you'll leak file handles.

Okay, thank you!

So I'm imagining something like this:

This is what I'm doing currently (after fixing it), do you think this will cause errors?

This version is more terse and does exactly what you want.