CMake 4.2 enables
Visual Studio 18 2026
generator.
The
FASTBuild
generator is added, and has impressive
benchmarks.
FASTBuild supports more modern platforms including Windows, versus
ccache
or distcc.
ExternalProject
added options to modify environment variables at each step by CONFIGURE_ENVIRONMENT_MODIFICATION, BUILD_ENVIRONMENT_MODIFICATION, INSTALL_ENVIRONMENT_MODIFICATION, TEST_ENVIRONMENT_MODIFICATION
CMake 4.1 project() added COMPAT_VERSION that propagates to subdirectories and can be queried for the top-level COMPAT_VERSION.
CMake 4.0 enables
CMAKE_LINK_WARNING_AS_ERROR
boolean option sets most compilers to error if a compile warning occurs, which is generally a good setting for CI systems.
target_link_libraries()
can use a LINKER: prefix to abstract options.
CMake
SARIF output
allows IDEs to better parse build status.
CTest
–interactive-debug-mode
is particularly useful on Windows to enable debug dumps and error popup windows.
CMake
3.31
warns if cmake_minimum_required() is
< 3.10.
TLS ≥ 1.2 is required by default for internet operations e.g. file(DOWNLOAD), ExternalProject, FetchContent, and similar.
file(ARCHIVE_CREATE)
gained a long-needed WORKING_DIRECTORY parameter that is essentially necessary to avoid machine-specific paths being embedded in the archive.
CMAKE_LINK_LIBRARIES_STRATEGY
allows specifying a strategy for ordering target direct link dependencies.
CMake
3.30
adds C++26 support.
CMAKE_TLS_VERIFY
environment variable was added to set TLS verification (true, false).
CMake 3.31 defaults CMAKE_TLS_VERIFY to on, where previously it was off.
CMake
3.29
adds
cmake_language(EXIT code)
to exit CMake script mode with a specific return code.
This is useful when using CMake as a platform-agnostic scripting language instead of shell script.
Environment variable
CMAKE_INSTALL_PREFIX
is used to set the default install prefix across projects–it can be overridden as typical by cmake -DCMAKE_INSTALL_PREFIX= option.
Target property
TEST_LAUNCHER
allows specifying a test launcher.
For MPI program this allows deduplicating or making more programmatic test runner scripts.
Linker information variables including
CMAKE__COMPILER_LINKER_ID
have been added to allow programmatic logic like setting target_link_options() based on the particular linker.
CMake
3.28
changes PATH behavior for
Windows find_{library,path,file}()
to no longer search PATH.
This may break some projects that rely on PATH for finding libraries.
The MSYS2-distributed CMake is patched to include PATH like earlier CMake, which can be confusing for CI etc. not using MSYS CMake with that patch.
Windows CI/user may need to specify CMAKE_PREFIX_PATH like
Support for C++20 modules is considerably improved and most users will want at least CMake 3.28 to make C++ modules usable.
Generator expressions $<IF> $<AND> $<OR> now short circuit.
Test properties now have a DIRECTORY parameter, useful for setting test parameters from the project’s top level CMakeLists.txt.
CMake 3.28.4 fixed a long-standing bug in Ninja Fortran targets that use include statements.
The brief listing of other, somewhat similar blogs is not in any way an endorsement or indication of curation.
It’s simply a notepad to remember what other blogs I might have once looked at that reminded me of content I’ve written or helped produce in the past.
Since blog addresses might change, discontinue, or go off the deep end, I just put the plain non-linked text.
Matlab or GNU Octave “.m” script functionality can be readily extended with Java code within the “.m” file.
Matlab or GNU Octave do not necessarily include Java from the factory installation.
Even if a
compatible JDK
is installed in Matlab or GNU Octave, Java can be disabled by
-nojvm.
Detect if JVM is available from within a Matlab script by:
GNU Octave also has a
JVM interface
that extends Octave functionality.
The same commands as for Matlab generally work in Octave via builtin functions javaMethod() and javaObject().
If Java isn’t available in Octave, see if no location or an out of date location to Java is set in the environment variable JAVA_HOME:
getenv('JAVA_HOME')
Currently GNU Octave doesn’t have a way to disable Java from the command line like Matlab’s -nojvm option.
One might try setting environment variable JAVA_HOME to an invalid location before starting Octave to simulate no JVM.
GCC has broad support of modern standards on a very wide range of computing platforms.
GCC is competitive in build time and runtime with vendor-specialized compilers. These may offer vendor-specific capabilities not available in general compilers like GCC.
Intel
oneAPI compilers
are free to use for any user, supporting Linux and Windows computers with x86-64 CPU including from AMD, Intel, etc.
Intel
oneAPI components
like MKL, IPP, and TBB are available at no cost.
Intel
MPI Library
implements the MPICH specification for massively parallel computation across discrete computing nodes.
LLVM Clang C and C++ compilers join with the
Flang
Fortran compiler using modern C++ internals and robust industry support.
LLVM is known for performance, correctness, and prompt implementation of new language standards.
LLVM has powerful associated tooling like clang-format, clang-tidy, and sanitizers.
It is generally important to ensure that a project builds with both LLVM and GCC for better portability.
AMD
AOCC LLVM compiler
is tuned for
AMD CPUs.
AOCC works on non-AMD CPUs but is generally only useful for those with HPC/AI workloads on AMD CPUs, as it typically uses LLVM releases that are a few versions behind the latest.
AMD GPUs are the focus of the
ROCm software stack,
which includes the HIP C++ language and ROCm Fortran compiler.
Currently the
ROCm Fortran Next Gen
compiler is in early development and is intended for advanced users who need to use AMD GPUs with Fortran or C/C++ code.
NVIDIA
HPC SDK
is free to use and works on a variety of desktop CPUs including x86-64, OpenPOWER, and ARM.
A key feature of the NVIDIA compilers is intrinsic support for
CUDA Fortran,
enabling offloading computationally intensive Fortran code to NVIDIA GPUs.
NVIDIA HPC SDK includes specialized tools for NVIDIA Nsight profiling, debugging, and optimizing HPC applications on NVIDIA platforms.
Unlike the other compilers mentioned above, IBM OpenXL LLVM-based compilers are specifically designed for POWER CPUs, such as ppc64le.
Consequently, IBM OpenXL compilers do not work with typical x86-based computers.
The IBM OpenXL Fortran compiler features extensive optimization capabilities specifically for POWER CPU architecture.
It supports OpenMP for parallel processing and can auto-vectorize code for POWER vector units (VSX, VMX).
The compiler includes built-in support for IBM MASS (Mathematical Acceleration Subsystem) libraries and optimization reports to help developers tune code performance.
OpenXL compilers include hardware-specific optimizations for POWER CPUs and support for IBM-specific operating systems like AIX.
DOSBox-X
uses configuration file dosbox-x.conf for numerous parameters, including for serial port access via USB to serial adapters.
First, identify the device ID for USB to serial adapter:
Linux: ls /dev/ttyUSB* or ls /dev/ttyACM*
macOS: ls /dev/cu.* or ls /dev/tty.*
On Linux, allow non-sudo serial port access by adding the username to “plugdev” and “dialout” groups, then logout and login.
adduser $(whoami) plugdev dialout
Locate
and edit dosbox-x.conf.
Under [serial] modify the serial1=dummy according to the device found above.
For example on Linux, if the device is at “/dev/ttyUSB0”:
serial1=directserial realport:ttyUSB0
DOSBox-X would present the serial device on COM1 as seen from within DOSBox-X.
Matlab Startup Accelerator
for Windows is intended to speed up Matlab startup.
However, by default Matlab disables the Startup Accelerator on many systems.
To check if Matlab Startup Accelerator is enabled on a Windows system, check Windows Task Scheduler by the snapin:
taskschd.msc
Look under Task Scheduler Library for a task named Matlab R20XXx Startup Accelerator (where 20XXx is the Matlab release).
Consider disabling it to avoid needless resource usage if it is not already disabled.
On modern computers with SSD drives, Matlab startup is already fast.
Python 3.15 added
os.statx()
function that provides access to the high performance
statx()
system call available on Linux kernels 4.11 and newer with glibc 2.28 and newer.
The Ffilesystem library uses statx() by default if available on Linux.
Having os.statx() available in Python allows for easier prototyping of features that might use statx().
See
Ffilesystem
statx() code for usage examples with graceful fallback to stat().
The
Minicom
serial communication program allows connecting to devices over the serial port and transferring files using XMODEM, YMODEM, or ZMODEM protocols–PuTTY can’t currently do file transfer over serial links.
Install Minicom like:
macOS: brew install minicom lrzsz (lrzsz is for file transfer)
Linux: apt install minicom lrzsz (or similar for relevant package manager)
Windows: use WSL (Windows Subsystem for Linux)
On macOS in particular for file transfers, first be sure that Minicom can find the sz, rz commands by looking under Esc-O for “File transfer protocols”. It may be necessary to point to lsz and lrz installed by
Homebrew.
As noted in the documentation for Minicom, it is possible to save profiles.
To just use a device directory, set command line options like:
minicom -D /dev/tty.usbserial-0001 -b 115200
List the USB-serial adapter ports:
Linux: ls /dev/ttyUSB*
macOS: ls /dev/tty.usbserial*
To send a file from Minicom once the device is ready, press Esc-Z and then S for send, select the appropriate protocol.
Concurrency is built into Python via
asyncio.
AsyncIO generators are implemented with yield much like synchronous generators.
async for
also simplifies expression of asynchronous for loops.
As in Julia, the expression of asynchronous structures in Python does not implement concurrent execution.
Concurrent execution in Python is governed by collections of tasks or futures such as
asyncio.gather
and initiated by a runner such as
asyncio.run.
Debugging asyncio code is facilitated by
introspection
built into Python.
AsyncIO
subprocess
may need specific
asyncio loop configuration.
The options needed are not the same for every project, depending on the asynchronous functions used.
Matlab can call user Python modules via the
Matlab external language interface.
For concurrent Python modules using asyncio, you may need to create a shim function to allow Matlab to call the Python module.
Anaconda Python works fine from Matlab.
Configure Matlab for Python: Matlab is designed to work with specific Python versions for each Matlab version.
Matlab will not specifically tell you when you’re using an incompatible Python version, but you may get unstable operation or errors.
pyenv
Python environment manager configures the Python interface.
The Python executable is found from Terminal / Command Prompt:
python -c "import sys; print(sys.executable)"
For Windows, it may be like C:/Miniconda3/python.exe and for macOS / Linux it may be like ~/Miniconda3/python.
This Matlab command is persistent–Matlab remembers this Python choice even after restarting Matlab.
pyenv(Version='C:/Miniconda3/python.exe')
Verify Matlab → Python config by typing pyenv from within Matlab.
If only Python standard library modules work, and even commonly used modules like Numpy will not work, errors may occur from Matlab like:
py.numpy.arange(1)
Unable to resolve the name py.numpy.arange
Try diagnosing the issue from Matlab like:
pyenv
help('py.numpy')
Verify the desired Python install is being used, and that there isn’t an error as below.
If such an error occurs, check the PATH as seen by Matlab by:
getenv('PATH')
If the Python directories are missing, this may be why user modules are not importable in Python.
These paths can be added to Matlab.
On Windows, the necessary path is like ~/Miniconda3/Library/bin, the directory with lots of *.{so,dylib,dll} files and the fix is like:
The install path on non-Windows OS is like ~/miniconda3/lib.
This command can be issued after a Python command was attempted, it’s not necessary to restart Matlab.
Note that this needs to be in PATH, since PYTHONPATH does not help in this case.
This change does not persist across Matlab sessions.
If it works for you, put it in the
~/Documents/MATLAB/startup.m
file to have it reload each time Matlab is started.
This Numpy problem is fixed by the procedure above:
problem in numpy - ImportError:
IMPORTANT: PLEASE READ THIS FOR ADVICE ON HOW TO SOLVE THIS ISSUE!
Importing the numpy c-extensions failed.
- Try uninstalling and reinstalling numpy.
- If you have already done that, then:
1. Check that you expected to use Python3.7 from "C:/Miniconda3/python.exe",
and that you have no directories in your PATH or PYTHONPATH that can
interfere with the Python and numpy version "1.17.4" you're trying to use.
2. If (1) looks fine, you can open a new issue at
https://github.com/numpy/numpy/issues. Please include details on:
- how you installed Python
- how you installed numpy
- your operating system
- whether or not you have multiple versions of Python installed
- if you built from source, your compiler versions and ideally a build log
- If you're working with a numpy git repository, try `git clean -xdf`
(removes all files not under version control) and rebuild numpy.
Note: this error has many possible causes, so please don't comment on
an existing issue about this - open a new one instead.
Original error was: DLL load failed: The specified module could not be found.
Python module import is implicit in the Matlab → Python module function call. There is no need to import numpy etc. from Matlab
Python executable choice persists across Matlab sessions–Matlab “remembers” even after you restart Matlab or reboot the computer.
editing imported Python module code requires restarting Matlab to take effect.