v4.2.0

Enhancements

• #614, #615 - Audio updates and cleanup - @a740g
  • Adds sample rate support to _SNDNEW.
  • Moves all information, warning, and trace logs to use our new logging features.
• #626 - Support for using System MinGW on Windows - @a740g
  • Adds support for using a system-installed compiler on Windows, such as a pre-existing MinGW toolchain (from MSYS2, Winget etc.). This can be helpful in situations where the user either cannot or prefers not to use the compiler bundled with QB64-PE.
  • This feature is available exclusively on Windows, a respective checkbox for switching is shown in the "Compiler Settings..." dialog under the "Options" menu, a clear warning is displayed when this option is enabled to ensure users understand the implications.
• #630, #634 - IDE Help System improvements - @RhoSigma-QB64
  • Added/fixed some Unicode char replacements.
  • The "Update All Pages" download process now excludes the mostly static _GL command pages, i.e. 337 files less to fetch, which saves us 2-3 minutes time.
• #635 - Add Windows on ARM support, closing issue #313 - @a740g
  • Enables full native support for QB64-PE on ARM-based devices, such as Windows devices using Qualcomm Snapdragon and macOS systems with Apple Silicon SoCs.
  • Eliminates reliance on runtime emulation (e.g. the Windows Prism x86 translation layer), enabling significantly faster startup times and improved execution speed.
  • The pre-compiler will set the _ARM_ flag and the _OS$ function return contains the string "[ARM]" when using QB64-PE on the ARM architecture.
  • On ARM architectures, the _FLOAT type is not natively supported due to the absence of x86-style 80-bit x87 FPU. This may lead to compatibility issues or reduced precision. Consider using DOUBLE or SINGLE for cross-platform consistency.
  • The VAL function previously returned values as _FLOAT by default. Users can now explicitly specify the target numeric type such as SINGLE, DOUBLE, or an integer type via the optional numericalType parameter, improving type control and cross-architecture compatibility.
• #641 - Adjustments to Windows set-up scripts - @SteveMcNeill
  • Replacing wmic with PowerShell equivalent.
• #642 - New dialog for easy logging configuration, closing issue #576 - @RhoSigma-QB64
  • Access the dialog via the menu "Run -> Configure logging...".
    

Library Updates

• #635 - Update libcurl to v8.14.0 - @a740g
• #640 - Update libxmp-lite to v4.6.2 - @a740g
  • Also, the clip, miniz, tsf, qoa, qoi source files were updated to the latest releases.

Bug Fixes

• #615 - Fix auto-formatting of _MEMPUT AS TYPE & _MEMFILL AS TYPE - @RhoSigma-QB64, @a740g
  • The case of the given variable type names was not adjusted according to the active keyword formatting (UPPER/CaMeL/lower).
• #616 - Fix _MOUSEX and _MOUSEY when the mouse pointer is hidden on macOS - @a740g
  • This partially reverses the approach taken in #602. Using CGAssociateMouseAndMouseCursorPosition() was a mistake - it fixed _MOUSEMOVEMENTX and _MOUSEMOVEMENTY for FPS-style camera movement but broke _MOUSEX and _MOUSEY by preventing GLUT from triggering the glutMotionFunc callback while the mouse pointer was hidden.
  • To address this, we now use CGEventSourceSetLocalEventsSuppressionInterval() to work around macOS’s 0.25-second event suppression. However, this means _MOUSEMOVEMENTX and _MOUSEMOVEMENTY remain broken because macOS generates artificial movement events when glutWarpPointer is used.
  • For now, this is an acceptable trade-off. A proper long-term fix depends on implementing #608.
  • Many thanks to NakedApe from the QB64-PE forum for pointing out the issue. See this.
• #627 - Fix IDE error when loading large source file - @FellippeHeitor
  • The IDE would crash when a large source file was loaded and the syntax highlighter had to build the list of sub/functions on its own, due to errors in the loaded file preventing compilation from going through.

Full Changelog: v4.1.0...v4.2.0