Memory leak on macOS

[benorei]

I installed tag studio on my laptop, played around for a bit, then I closed my laptop and when I came back a few hours later, my computer had run out of memory because tag studio was trying to use 114 GB of memory.

[CyanVoxel]

Well this doesn't look very good...
Could I get some more build/system information from you?

• TagStudio Version + Build Type (Git, macos_x86_64, macos_aarch64)
• macOS Version
• CPU model
• RAM amount
• What was the last thing you did in the program?

[xarvex]

Hi @benorei, can we get the information asked about so we can explore what's going on? As well, is this something you've encountered again at any point?

[benorei]

Yes, sorry.
I don't know what version - I have since uninstalled. It was the latest version at whatever day I made the post. I downloaded it from Github, and it would be macos_aarch64.
macOS Version - Sonoma 14.5
CPU model - Apple M2 Pro
RAM amount - 32 GB
Last thing I did in the program - Load a large amount of images into the viewer

[yedpodtrzitko]

While working on the sql backend, I most likely stumbled upon where the memleak is happening.
QtDriver.frame_dict is accumulating data with each library search and never clearing them.

this can be reproduced by printing memory usage of the object here:

TagStudio/tagstudio/src/qt/ts_qt.py

Line 1375 in ce87b11

self.frame_dict[query] = frames

install memory analyzer

$ pip install pympler

add into the code:

from pympler import asizeof
print(asizeof.asizeof(self.frame_dict))

now with each new search you will see that the memory size of frame_dict is increasing