This PR improves the safety and reliability of the deleteOldFiles()
function by:
Preventing symlink attacks
Uses std::filesystem::canonical() to resolve all paths, eliminating ./..
and symlinks
Verifies each file is actually inside the target directory before
deletion
Safer file operations
Explicitly skips non-regular files (directories/symlinks)
---------
Co-authored-by: Nik <werwolv98@gmail.com>
With an initial Window size of 1 by 1, the program aborts during launch
on my Debian 12 (bookworm) system running the Mate Desktop environment
on X11.
The exception from X is the following:
```
$ ./imhex
[18:36:33] [INFO] [main | Main] Welcome to ImHex 1.38.0.WIP!
[18:36:33] [INFO] [main | Main] Compiled using commit master@59c6d30
[18:36:33] [INFO] [main | Main] Running on Linux 6.1.0-37-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.1.140-1 (2025-05-22) (x86_64)
[18:36:33] [INFO] [main | Main] Linux distribution: Debian GNU/Linux 12 (bookworm). Version: 12
X Error of failed request: BadWindow (invalid Window parameter)
Major opcode of failed request: 18 (X_ChangeProperty)
Resource id in failed request: 0x5400009
Serial number of failed request: 152
Current serial number in output stream: 163
```
The culprit is in `hex::init::WindowSplash::initGLFW()`, specifically
the `glfwCreateWindow` call with a window size of 1 by 1, that's the
last call in the ImHex codebase, after that it's libglfw and libX11.
The issue can be worked around by inhibiting the splash screen using the
`IMHEX_SKIP_SPLASH_SCREEN=1` environment variable, setting the window to
be a size larger than 1 by 1 also seems to resolve the situation.
My hunch is that an underflow occurs somewhere with the smaller values,
but I did no further investigation into analysing the root cause.
----
Above the line is the commit message.
### Problem description
See https://github.com/WerWolv/ImHex/issues/2215, which I think is the
same problem, and commit description above. Edit; I did another search
in the issues, I wouldn't be surprised if #2233 is a duplicate of the
2215 issue.
### Implementation description
Please see the commit message.
Analysis was done by setting an X11 error handler, making that throw and
running it in a debugger. This is available in this dev commit;
4a24e16a2ef100a9417e572f6b16c4a95711e94e.
### Screenshots
No apparent glitches if I make the splash window stick around with
`return std::nullopt;` in the `loop()` function. I don't know if there's
an issue at the beginning, but given that the window is set to
transparent before it is created and resized I don't expect any.
Here's a compulsory screenshot of the splash screen over an editor;
<img width="1247" height="666" alt="image"
src="https://github.com/user-attachments/assets/8ce05565-f207-4134-892d-9c074c4c840c"
/>
### Additional things
> Anything else you would like to say
Yes! I don't use ImHex often, but it's really useful when I do, thank
you!
