GUI less dependent on MATLAB and MathWorks

Two recent developments with MathWorks:
- MATLAB shifting from java-based GUI to browser-based GUI
- relatively long service outage at MathWorks (10+ days now), preventing download of MATLAB 2019b runtime library (for the FEATool GUI, if not integrating directly with MATLAB as an application). UPDATE: MathWorks site allows download of the MATLAB 2019b runtime library now.

Are there concerns that MathWorks will cease support or availability of the MATLAB 2019b runtime library now that the java-based GUI is no longer being used?

Is this an opportunity to create a different type of GUI for FEATool; one that is not fully dependent on any single provider?

I am not keen on using a web-based interface for MATLAB. I am also not keen on remaining at MATLAB 2024b for a long period of time, due to potentially missing out on bug fixes and security patches.

I am inclined to replace MATLAB with other tools, such as GNU Octave and OpenModelica. While having FEATool Multiphysics integrated with MATLAB as an application can add some efficiency, I don't find this to be a requirement for my continued use.

We were really caught off guard by the complete removal of the Java GUI in the Matlab R2025a beta, as it had not at all been warned or communicated to us by the Mathworks when the Java GUI would be deprecated.

As such we have currently dropped working of new features, and instead working on a Python based frontend/GUI replacement, as we don't think the performance (and look) of the current Matlab Web UI is good enough for a professional App (it is odd that running an UI though Python via a Java-TCP bridge is significantly more performant and responsive that the current native Matlab implementation). In the long run the toolbox with be migrated to a Python based toolbox, that optionally can connect to a Matlab or Octave backend, with the initial release of the new UI slated for sometime late(er) this year.

Regarding Matlab versions, with respect to the FEATool toolbox there is typically very little gain from using newer Matlab versions. The built-in linear solvers (Tim Davis's SuiteSparse/UMFPACK) which do most of the computational work and unlikely to be improved significantly upon as the algorithms already pretty much max out CPU performance and system memory bandwidth. As such we typically recommend using Matlab 2019b as we have found it most stable with the external MUMPS linear solver (which is why it is also used as a base for the stand-alone version).

Wow - great! That's exciting to hear. Thanks for your prompt response and best wishes with your effort.