Compatibility

ALGLIB aims to be one of the most portable numerical analysis libraries. There is no reason it shouldn't work under any platform which is at least 32-bit and supports IEEE-compatible double precision floating point operations. However, ALGLIB was tested only with a limited number of platforms/compilers (see below). If your OS/CPU/compiler is not in the list, you still may compile ALGLIB yourself and manually execute tests for unit you want to use.

             OK
             Minor issues
             Major issues
             Not tested yet, planned in the near future
             Nonexistent combination of CPU/OS/compiler

Cross-platform tools
Language/compiler		Windows		Linux		OpenSolaris
		x86	x86_64	x86	x86_64	x86	x86_64
C#	Mono
FreePascal	FPC		major issues*
C++	Intel C++	not tested	not tested	not tested	not tested
	GCC					not tested	not tested
	Sun Studio			not tested	not tested	not tested	not tested
MPFR	GCC					not tested	not tested
	Sun Studio			not tested	not tested	not tested	not tested

* Compiler bug. FreePascal Compiler for Win64 includes standard library which loses precision for Exp() and, possibly, another functions, which causes ALGLIB to fail tests.

Windows-only tools
Language/compiler		Win32	Win64
C++	MSVC
	MinGW	minor issues*
	MinGW-W64	not tested	not tested
	Cygwin	not tested	not tested
C#	MS C#
Delphi	DCC32
MPFR	MSVC	not tested	not tested
VBA	VBA	occasionally tested**	occasionally tested**

* MinGW bug (standard distribution includes very old version of GCC). Program crashes when compiling with "-Os". Problem is classified as "minor" because it is easily detectable and can be cured by compilation with any other optimization settings.

** Because it is impossible to launch VBA programs from command line, unit testing is occasional and is done in manual mode.

This article is intended for personal use only.