Bug fixes for Massive UX improvements, Custom Disassemblers, MSVC support

[WerWolv]

Bug Fixes

• Fixed large memory allocations when printing error messages in some cases
• Fixed an issue that caused the pattern editor to shift to the left when clicking evaluate
• Fixed a crash when a custom disassembler returned no bytes
• Fixed incorrect native scaling value when using fractional scaling on Wayland
• Moving the ImHex window between two screens with different DPI on Windows works better now
• Fixed a crash when loading a project with a saved process memory provider
• Fixed closing one Selection View causing all other Selection Views to become unusable
• Fixed a crash when selecting an invalid font
• Fixed the pattern editor losing focus when auto evaluate is on

Previous Version

Features

• ImHex now properly supports HiDPI scaling on Windows, macOS and Linux!
• ImHex now uses the native menu bar on macOS
• You can now set separate fonts for the main UI, the Hex Editor and the Pattern Editor
• Added a new ARM64 AppImage release as well as a WebAssembly release to self-host ImHexWeb
• ImHex can now be fully compiled using MSVC and ClangCL on Windows! Huge thanks to @mrexodia
  • The main releases are still built with MinGW but this should hopefully make it easier to contribute to ImHex!
• We now provide build attestation for each build generated through our CI so you can verify exactly from which run an artifact has been built
• Added initial support for custom disassemblers
  • This lets you define disassemblers for custom instruction sets using a simple JSON format
• Added messaging support for macOS and Linux
  • This allows you to use ImHex as a command line tool on macOS and Linux to control the currently open ImHex instance
  • For example imhex --open "SomeFile.bin" --select 0x100 0x200 will open the SomeFile.bin file in the current instance and then select addresses 0x100 to 0x200
• Added Russian translation! Huge thanks to @Lemon4ksan
• Added a title bar backdrop color gradients and accent colors
• Added utf16 and utf32 string and char types
• Added option to randomize window title
  • In case some stupid anti-cheat is trying to detect ImHex in the future :)
• Added new --select, --pattern and --debug-mode subcommands
• Added support for OpenGL post processing shaders
  • This lets you load custom shaders now by placing a shader.vert and shader.frag file in one of the ImHex resource folders.
  • These shaders can apply any post processing to the rendered image
• Added New File option to the GNOME launcher and a --new cli option
• Added support for scanning binaries for UTF-8 strings
• Added visibility toggle to hide bookmark highlighting
• Added option to fit hex columns to screen width
• Added setting to disable hex editor highlights entirely
• Added a preview to the Edit -> Copy as options
• Added "Jump to address" option to data inspector row context menu
• Added option to change radix of numbers in hex editor view
• Added context menu and next/previous buttons to the data inspector
• Added option to move selection back to hex editor footer
• Added option to copy data as escaped string
• Added shortcut for Copy as -> Custom Encoding

Improvements

• Replaced the Windows and Linux app icon with the one from the Papyrus Icon Project!
• The DMG on macOS now looks a lot nicer than before
• Revamped the frame rate limiting system to make ImHex feel much less laggy by default and make it use less CPU resources
• ImHex now uses Jetbrains Mono as its default font instead of the ImGui default font
• The settings window can now be resized
• The ImHex window is now hidden on macOS when close button is pressed
• Allowed all highlights to overlap each other
• Allowed the favorite column in the pattern data view to be hidden
• Allowed command palette to be closed by clicking on the menu bar
• All installed content store items are now force updated after an ImHex update

Bug Fixes

• Fixed various issues with the auto updater
• RGBA8 data processor node not working correctly
• Very slow processing of large pattern console outputs
• Windows forwarder application not working when piping output
• Writing not working correctly through a provider view
• Crash on Linux if XDG_SESSION_TYPE is not set
• Crash when closing ImHex with one or more view providers open
• Uninitialized buffers in resize operation
• Exception being thrown while loading projects
• Crash when certain pattern language exceptions were thrown
• Pattern Editor shortcuts not working correctly on different languages (fix: fixes for issue #2084 #2085)
• Pattern Editor Find and Replace history (fix: find-replace history #2064)
• Crash in 3D Visualizer, improved error messages (fix: unreported crash on 3d-visualizer #2062)
• Disable bogus Keypad to function key conversions on macOS
• Empty regions in Intel Hex and Motorola SREC files not being displayed correctly
• Make sure provider switch buttons and close button don't overlap
• Hang when filtering for a large number of items in the pattern data view
• Off-by-one when calculating hashes of selected regions
• ImHex freezing on AMD GPUs when resizing
• Rendering issues with text editor in bookmark view
• Ctrl sometimes getting stuck when pressing ALT GR and other keys at the same time
• Missing endian setting in ARM64 disassembler
• Clicking past end of line in text editor putting cursor before last character

Pattern Language

• The evaluator now tries to keep as many patterns as it possibly can when an error occurs
• Added cursor positions to the error stack traces
• Exported global variables are now displayed in the output
• Added support for rvalue assignments inside of structs
• Fixed various issues when importing patterns as types
• Make decompress functions return number of read bytes
• Allow #pragma magic to index from the end of the data with negative addresses
• Fixed error spam when encountering an error in the Lexer

If you like my work, please consider supporting me on GitHub Sponsors, Patreon or PayPal. Thanks a lot!

This discussion was created from the release Bug fixes for Massive UX improvements, Custom Disassemblers, MSVC support.

[eloj]

Both the 1.37.0 and 1.37.1 releases seems to invoke some sort of memory error on startup for me (Linux Mint/X11 (nvidia 570.86.16), AppImage versions), leading the compiler stack protector to get invoked:

[12:11:11] [INFO]  [main | Loading fonts]      Task 'Loading fonts' finished successfully in 778 ms
[12:11:11] [INFO]  [main | Init Tasks]         ImHex fully started in 2239ms
*** stack smashing detected ***: terminated
*** stack smashing detected ***: terminated
[12:11:12] [INFO]  [main | Main]               Switching to workspace 'Default'

ImHex still starts, but it leaves two core files behind:

$ file core.sh.1481*
core.sh.14813: ELF 64-bit LSB core file, x86-64, version 1 (SYSV), SVR4-style, from 'sh -c -- dbus-send --session --print-reply --dest=org.freedesktop.portal.Deskto', real uid: 1000, effective uid: 1000, real gid: 1000, effective gid: 1000, execfn: '/bin/sh', platform: 'x86_64'
core.sh.14814: ELF 64-bit LSB core file, x86-64, version 1 (SYSV), SVR4-style, from 'sh -c -- dbus-send --session --print-reply --dest=org.freedesktop.portal.Deskto', real uid: 1000, effective uid: 1000, real gid: 1000, effective gid: 1000, execfn: '/bin/sh', platform: 'x86_64'

The most similar issue I saw was #2089, but I'm not running Wayland.

[WerWolv]

Interesting! That code has not been touched in a long time, It's the code that asks your desktop environment over dbus what theme color you have chosen in your settings. Are you sure it's not an issue with your setup somehow? Since it's the dbus-send process that's crashing, not ImHex

[eloj]

I wouldn't rule it out, but I also can't reproduce it just by sending the query manually. If dbus was completely broken on my system, then nothing would work.

$ cat ./testdbus.sh 
dbus-send --session --print-reply --dest=org.freedesktop.portal.Desktop /org/freedesktop/portal/desktop org.freedesktop.portal.Settings.Read string:'org.freedesktop.appearance' string:'color-scheme' 2>&1
$ sh -c ./testdbus.sh
method return time=1739880874.305971 sender=:1.104 -> destination=:1.1595 serial=569 reply_serial=2
   variant       variant          uint32 0

Oh, well. I'm not super-invested in tracking this down, but now you know.