doc

Author: scivision

+stdlib/Backend.m

@@ -1,3 +1,13 @@
+%% BACKEND auto-selection of function backend per-function
+% the user can specify a backend, or else this class automatically
+% determines suitable backends for the system.
+%
+% * 'sys' is usually the slowest, but most compatible.
+% * 'python' or 'dotnet' are among the fastest, but not always available
+% * 'java' and 'perl' are usually available and medium speed
+% * 'native' uses the latest Matlab syntax available
+% * 'legacy' is Matlab syntax for older Matlab versions. Sometimes legacy is faster than native, but native is more robust/general.
+
 classdef Backend < matlab.mixin.SetGet
 
 properties (Constant)

+stdlib/perl_exe.m

@@ -1,3 +1,6 @@
+%% PERL_EXE get the location of the Perl executable used by Matlab
+% this is cached for speed
+
 function exe = perl_exe()
 
 persistent perle
@@ -16,4 +19,4 @@
   perle = r;
 end
 
-end
+end

+stdlib/perl_version.m

@@ -1,4 +1,5 @@
 %% PERL_VERSION get the Perl version used by MATLAB
+% cached for speed
 %
 %%% Output
 % * v: 1x3 vector of major, minor, micro version e.g. Perl 5.32.1 = [5, 32, 1]

Readme.md

@@ -41,7 +41,8 @@ Normally the user will not specify the backend as we prioritize faster backends.
 * [.NET](https://www.mathworks.com/help/matlab/call-net-from-matlab.html)
   * Windows: all supported Matlab releases
   * Linux / macOS: R2024b and newer
-* [Java](https://www.mathworks.com/help/matlab/using-java-libraries-in-matlab.html): all supported Matlab releases. A few Java functions are Linux / macOS only, but have other backends available.
+* [Java](https://www.mathworks.com/help/matlab/using-java-libraries-in-matlab.html): all supported Matlab releases
+* [Perl](https://www.mathworks.com/help/matlab/ref/perl.html):  all supported Matlab releases. This uses a system() call to Perl, which is bundled with Matlab on Windows and generally available on Linux and macOS
 * [Python](https://www.mathworks.com/help/matlab/call-python-libraries.html): Matlab R2022b and newer
 * System shell calls: all supported Matlab releases. As a backup when the platform doesn't have the primary (faster) methods available, the system shell can be called for some functions.