Depending on application requirements, the memory footprint of MATLAB^®Runtime is now decreased by up to 60% when compared to previous releases.

Excel^®add-ins created using a 64-bit installation ofMATLAB Compiler™can be installed and used by 32-bit installations of Microsoft^®Office.

Use thedatastorefunction to create anImageDatastore, which can import data from image collections that are too large to fit in memory.ImageDatastoreis a compatibledatastorefor use withmapreduce.

• Functionality to create Excel add-in components for Microsoft Excel desktop applications has moved to MATLAB Compiler.

• Functionality to create C/C++ shared libraries has moved toMATLAB Compiler SDK™.

• Functionality to create CTF archive files for use withMATLAB Production Server™has moved toMATLAB Compiler SDK.

MATLAB Compileranalyzes your MATLAB code for use of graphics or a GPU and defines and calculates optimized MATLAB Runtime. A MATLAB Runtime specific for each application is available for download from the web or included in the package.

MATLAB Compilernow supports MATLAB objects as output from the map and reduce functions for Hadoop^®Integration.

MATLAB Compilerincludes a MapReduce compiler app that packages MapReduce applications for deployment against Hadoop. The applications can be compiled into either standalone applications or libraries for use in developing other applications.

MATLAB Compileruses a new dependency resolution function that is faster.

Compatibility Considerations

The new dependency resolution function requires that p-code being used in a compiled function has an associated.mfile containing the original source for the p-code.

The compiler apps now scan your code and the installed support packages to determine what needs to be packaged. The compiler will package all required support packages for distribution.

TheMATLAB Compilerapps generate a platform specific installer that will automatically download and install the proper MATLAB Compiler Runtime if it is needed.

TheMATLAB Compilerapps use a toolstrip for commonly used interface components and required input fields.

TheMATLAB Compilerapps have fields for supplying custom graphics, icons, developer profile details, and version numbers to the generated installer.

MATLAB Compilernow supports C and C++ shared libraries generation for Mac OS X, using Apple Xcode compiler, version 4.2.

For more information on using Mac withMATLAB Compiler, seeWriting Applications for Mac or Linux.

TheMATLAB Compiler Runtime (MCR)is available for downloading from the Web to simplify the distribution of your applications or components created with theMATLAB Compiler.

Direct your end users to theMATLAB Compiler Runtimepage at//www.tianjin-qmedu.com/products/compiler/mcr/to download the MCR, as opposed to redistributing or packaging it with your applications or components.

Thelcccompiler that ships with Windows^®will no longer be supported as of R2013a.

For release R2012b, selecting thelcccompiler usingmbuild -setupresults in a warning, though the compiler is available. In R2013a,lccwill no longer appear in the list of available compilers.

TheMATLAB Compiler Runtime (MCR)is available for downloading from the Web to simplify the distribution of your applications or components created with theMATLAB Compiler.

Direct your end users to theMATLAB Compilerproduct pageto download the MCR, as opposed to redistributing or packaging it with your applications or components.

MATLAB Compilernow supports Linux^®gcc4.4.xcompilers for both 32-bit and 64-bit architectures.

See//www.tianjin-qmedu.com/support/compilers/current_release/for a complete up-to-date list of supported compilers and software.

An example demonstrating how to deploy a graphical MATLAB function with the Mac OS X Cocoa development layer, is available.

The function is compiled to a shared library target and then integrated into a deployed application using Apple Xcode.

SeeWriting Applications for Mac or Linuxfor more details.

Parallel Computing Toolbox™ configurations are now referred to asparallel profiles, as of R2012a.

SeeProfiles in Compiled Applicationsin theParallel Computing Toolbox Release Notesfor complete information.

Compatibility Considerations

The older MCR User Data Interface argumentParallelConfigurationFileand MAT files will still be supported until Parallel Computing Toolbox documentation states otherwise, but note that if you use theParallelConfigurationFileargument, you must supply a configuration.matfile, rather than a profile.settingsfile. If theParallelProfileargument is used, a configuration.matfile or a profile.settingsfile can be used.

The following changes toMicrosoft Visual Studio^®support apply for R2012a and R2012b.

Thelcccompiler that ships with Windows will no longer be supported as of R2013a.

For releases R2012a and R2012b, selecting thelcccompiler usingmbuild -setupresults in a warning, though the compiler is available. For R2013a,lccwill no longer appear in the list of available compilers.

The Application Deployment products now feature faster installation of the MATLAB Compiler Runtime (MCR) through improved MCR Installer technology. In addition to improved performance, the new installer provides a common interface for all platforms.

MATLAB Compilergenerated executables and components now benefit from better performance when used with Parallel Computing Toolbox applications. The improvement is due to support of the Graphics Processing Unit (GPU) feature.

For more information, seeGPU Computingin theParallel Computing Toolbox User's Guide, as well as the Parallel Computing Toolbox examples in the deployment user's guides for additional information about compiling your deployable application with GPU.

MATLAB Compilerand the builders now support local workers in Parallel Computing Toolbox applications.

For information about this feature, seeUse a Local Clusterfor general information on Local Workers. Also, see theParallel Computing Toolbox User's Guidefor information about working with compilable PCT applications.

MATLAB Compilernow provides error messages in pop-up windows for Windows and Macintosh graphical applications. You do not need to run a console in order to view these messages.

Effective with this release,Microsoft Visual StudioExpress 2010 compilers (32- and 64-bit versions) are now supported.Microsoft Visual StudioExpress 2008 compilers (32- and 64-bit versions), which have been dropped from support. SeeInstallation and Configurationin theMATLAB CompilerUser's Guidefor information about installing and configuring them. See//www.tianjin-qmedu.com/support/compilers/current_release/for an up-to-date list of supported compilers.

This release adds support forMicrosoft Visual Studiosoftware on both 32-bit and 64-bit systems. See theInstallation and Configurationchapter in theMATLAB CompilerUser's Guidefor more information.

This feature lets you add a link to an MCR Installer residing on a local area network. Adding this link allows you to invoke the installer over the network, as opposed to copying the installer physically into each deployable package.

SeePackaging (Optional)in theMATLAB Compiler User's Guide, or in your respective Builder product User's Guide, for more details.

You no longer need to specify a third-party compiler withmbuild -setupwhen building a standalone target. This is the case for both standalones and Windows standalones (standalone applications that suppress the DOS command window when running). For more information about whenmbuild -setupis not needed, seeWhen Not to Use mbuild -setup.

In previous releases, you could manually manage cache formccon Linux systems.

As of R2010b, caching formccis automatically managed and tuned. No user intervention is required.

In previous releases, the workflow described in the procedure "Mixing MATLAB Files and C or C++ Files" detailed how to mix MATLAB and C or C++ source files. This workflow has been replaced with a more straight-forward approach for release R2010b.

Compatibility Considerations

过时的工作流(记录之前运行R2010b) now results in a warning.

SeeCombining Your MATLAB and C/C++ Codefor details on the recommended replacement workflow.

As of R2011a, themcc -Foption will no longer process Deploytool Tool project files passed in or generate project files when invoked with command line arguments.

Compatibility Considerations

In R2010b, a warning will appear when themcc -Fcommand is issued. In R2011a, the command will generate an error.

Instead of usingmcc -F, seeUsing the Deployment Tool from the Command Linein the User's Guide for replacement functionality.

Use of the%#externalpragma to combine C/C++ code with a MATLAB application has been deprecated for standalone targets in release R2010b.

Compatibility Considerations

Using the%#externalpragma in release R2010b results in a warning. In a future release, usage will result in an error.

SeeInterfacing MATLAB Code to C/C++ Codefor details of the replacement workflow.

Beginning in release R2010b, the Deployment Tool will store projects in a new format.

Compatibility Considerations

If you open a supported legacy project file withdeploytool, you will automatically be prompted to save it in the new format.

If your legacy project files are in a nonuser-customized format, there is no action you need to take beyond converting your projects to the new format, when prompted.

Intel Macintosh 32-Bit systems, also known as Maci, are no longer supported as of release R2010b.

The software still supports Intel Macintosh 64-Bit systems, also known as Maci 64.

Compatibility Considerations

所有电脑包由部署连续cts are designed to run on 64-bit architectures and specific releases of the Apple operating system. Incompatibilities will result in an error from the Apple operating system.

Macintosh 64-bit applications can now be launched conveniently from the desktop. SeeUsing MATLAB Compiler on Mac or Linuxfor details about the Mac Application Launcher and other guidelines for UNIX, Linux, and Mac users.

Watch a video about deploying applications usingMATLAB Compiler.

This release provides support for MATLAB sparse matrices via a C++ API. SeeStatic Factory Methods for Sparse Arraysfor a listing of capable classes and methods.

You can now invoke the Deployment Tool GUI from the command line. SeeUsing the Deployment Tool from the Command Linefor more information.

In MATLAB, settinglinkdata oncan be used to link data between multiple plots in the same figure. This feature is not available for standalone applications run in deployed mode.

As described in theMATLAB Release Notes, the version number of the MATLAB Compiler Runtime (MCR) in release R2009bSP1 is different from the MCR version number in release R2009b.

The new Deployment Tool (deploytool) interface features intuitive task-based navigation, a cancellable progress dialog, fast loading of previously-created projects, and ability to add supporting files as folders. The GUI also features new context sensitive help.

Console applications can now display a user-customizable message indicating that the MATLAB Compiler Runtime (MCR) is in the process of starting up. SeeDisplaying MCR Initialization Start-Up and Completion Messages For Usersfor more information.

The same memory shielding process available in MATLAB is now available for deployed applications. The process reserves memory to ensure resource-intensive applications are allocated proper run-time resources. SeeReserving Memory for Deployed Applications with MATLAB Memory Shieldingfor more information.

Support for Macintosh^®64-bit processors is now available.

As of this release, deployment projects are being created in a new file format.

Compatibility Considerations

If you have projects that were created with the Deployment Tool prior to R2009b, those projects will continue to work. However, projects created or changed in R2009b cannot be opened in previous versions of Deployment Tool (deploytool). Recreate old projects with Deployment Tool to ensure they function properly.

The pathmatlab\bin\archhas been removed for Windows and replaced bymatlabroot\runtime\win32|win64. You may be required to reconfigure your deployment target computers. SeeSet MCR Paths for Development and TestingandSet MCR Paths for Run-Time Deploymentfor more information.

For more information on this change in support that will occur in R2010a, seeFunctionality Being Removedin theMATLAB Release Notes.

MCR copying and transfer time has been improved by 50% following a reduction in the file's contents. The change will not reduce processor time or memory consumption since the deleted files were not loaded into RAM.

You can now compile protected MATLAB files (Pcode files or "P-files"). For more information about Pcode and P-files, see theMATLAB Function Reference.

New functions have been introduced that return data about MCR state, such as whether the MCR has been properly initialized or whether the MCR was launched with a JVM™ instance enabled. For more information, seeRetrieving MCR Attributesin the user guide.

Microsoft Visual StudioExpress 2008 compilers (32- and 64-bit versions) are now supported instead ofMicrosoft Visual StudioExpress 2005 compilers (32- and 64-bit versions), which have been dropped from support. SeeInstallation and Configurationin theMATLAB CompilerUser's Guide for a complete list of supported compilers and information about installing and configuring them.

An enhanced version of thereadme.txtfile, which is generated with each successful compilation, is now customized to include specific instructions aboutMATLAB Compilerdeployment requirements.

You can now compile MATLAB applications that use the Parallel Computing Toolbox. Resulting executables and components can scale to multicore and multiprocessing environments using theMATLAB Distributed Computing Server™. For more information, seeImproving Data Access Using the MCR User Data Interfacein theMATLAB CompilerUser's Guide for more information.

It is now possible to share data between an MCR instance, the MATLAB code running on that MCR, and the wrapper code that created the MCR, by implementing two MATLAB functions and four external C functions in a new API that may be called from within deployed application wrapper code. Using these functions may potentially improve performance and promote efficient use of computing resources. SeeImproving Data Access Using the MCR User Data Interfacein theMATLAB CompilerUser's Guide for more information.

TheMX_COMPAT_32_OFFvariable, which allowed 32-bit addressing ofmwArrays, is no longer supported. As in R2008a, the default is 64-bit (large array support). See "Addressing mwArrays Above the 2 GB Limit" in theMATLAB CompilerUser's Guide for more information.

Support is now available forMicrosoft Visual Studio2008, 32-bit and 64-bit editions.

As of R2008b, running a figure-generating application or printing with the-nojvmoption results in a warning message.

Compatibility Considerations

In some cases, figure rendering may succeed, and in other cases it may not. Similarly, various MATLAB graphics functions dependent on Java^®, such as graphics passed with-R -nojvm, will not result in displayable graphics. In a future release, the-nojvmoption will no longer support figure-generating or printing and will be removed.

If the same class name is used in two or more separate MCR instances within the same process, the object will not work correctly. This bug impacts all component based targets: C/C++ shared libraries, COM/.NET targets, and Java targets. This bug does not affect standalone executables. This is due to a bug in the MATLAB Object System and is being addressed.

In R2007b, theMX_COMPAT_32_OFFvariable allowed for 64-bit support. In this release, 64-bit is the default setting.

Compatibility Considerations

If you still require explicit 32-bit execution, see "Addressing mwArrays Above the 2 GB Limit" in theMATLAB CompilerUser's Guide for more information.

As of R2008a, CTF data is now automatically embedded directly in the C/C++, main and Winmain, shared libraries and standalones by default for convenient deployment of applications. In order to override this default functionality, you must compile with the-Coption. See "MCR Component Cache and CTF Archive Embedding" in theMATLAB CompilerUser's Guidefor more information.

Support has been added for the compiler included withMicrosoft Visual Studio2008. See "System Requirements" in theMATLAB CompilerUser's Guidefor more details.

Starting with R2008a, replace all occurrences of#include mclmcr.hwith# include mclmcrrt.h.

Compatibility Considerations

If the library header created byMATLAB Compilersoftware is included in the external code, neither of these files need to be included, since the library header includes the correct file.

If you distribute a MEX-file, an engine application, or a MAT-file application built with the Visual Studio 2008 compiler, you must provide the Visual C++^®run-time libraries. These files are required to run applications developed with Visual C++ on a computer that does not have Visual C++ 2008 installed. For information on locating theMicrosoft Visual C++2008 Redistributable Package (x86), containingvcredist_x86.exeandvcredist_x64.exe, consult your Microsoft documentation.

In R2008a, the default definition ofMX_COMPAT_32has been removed, and large array support is now the default for both C and C++ code. This new default may, in some cases, cause compiler warnings and errors. You can defineMX_COMPAT_32in yourmbuildstep to return to the previous default behavior.

Code compiled withMX_COMPAT_32isnot64-bit aware. In addition,MX_COMPAT_32controls the behavior of some type definitions. For instance, whenMX_COMPAT_32is defined,mwSizeandmwIndexare defined toints. WhenMX_COMPAT_32is not defined,mwSizeandmwIndexare defined tosize_t. This can lead to compiler warnings and errors with respect to signed and unsigned mismatches.

For information about expected behavior of this feature in R2007b, see "Addressing mwArrays Above the 2 GB Limit" in theMATLAB CompilerUser's Guide for more information..

This feature changes how theMWArrayC++ Library functions are implemented. For more details, see "Addressing mwArrays Above the 2 GB Limit" in theMATLAB CompilerUser's Guide.

Borland compilers are no longer supported as of this release.

The MATLAB Component Runtime has been renamed to the MATLAB Compiler Runtime.

A new publication, theMATLAB Application Deployment Web Example Guide, is now available from theMATLAB Compiler,MATLAB BuilderNE, andMATLAB BuilderJAroadmap pages.

The guide provides full examples of common tasks performed by the MATLAB programmer, IT specialist, and others who play significant roles in deploying MATLAB applications to the Web.

As of R2008a, the LCC compiler is more strict in enforcingboolreturn types from C andvoidreturns from C++. For more information, seeCode Proper Return Types From C and C++ Methodsin theMATLAB CompilerUser's Guide.

Changes introduced in this release now require that the header file mclmcr.h always be included after mclmcrrt.h in the preprocessing stage.

Compatibility Considerations

This change introduces a potential compatibility problem in subsequent releases ofMATLAB Compiler. See theMathWorks support Web sitefor further information.

Support has been added for the Microsoft Platform SDK Compiler. SeeSystem Requirementsin theMATLAB CompilerUser's Guidedocumentation for more details.

Project files can now be enabled for use with other computers since project files now use relative paths. SeeUsing Relative Paths with Project Filesin theMATLAB CompilerUser's Guidedocumentation for more details.

In past releases, you needed to includeMCRInstaller.zipin your packaged application (created by running thebuildmcrcommand). Now you must run the following files, which trigger self-extracting archives, that replace the functionality previously provided byMCRInstaller.zip. These files ship withMATLAB Compiler:

Since there is no longer a need to createMCRInstaller.zip,buildmcris no longer supported.

Typemcrormcrinstallerat the MATLAB command prompt for a list of all available MCR installers, compatible platforms, version numbers, and other information. In addition, typinghelp mcrorhelp mcrinstallerat the command prompt will provide further details and assistance.

SeeDistribute MATLAB Code Using the MATLAB Compiler Runtime (MCR)for more information.

As of R2007b, you can addressMWArrays above the 2 GB limit. This is enabled by definingMX_COMPAT_32_OFFin yourmbuildstep. This feature will be the default as of R2008a.

This feature changes how theMWArrayC++ Library functions are implemented. For more details, see "Addressing mwArrays Above the 2 GB Limit" in theMATLAB CompilerUser's Guidedocumentation.

This newmccoption suppresses the appearance of the MS-DOS command window when generating a standalone application. Use-ein place of the-moption. See-e Suppress MS-DOS Command Windowin theMATLAB CompilerUser's Guidedocumentation.

Borland compilers are supported in Version 7.5 (R2007b), but will not be supported in a future version of MATLAB. Please prepare and plan accordingly.

The following functions were removed in R2007b:

Support has been added for the following operating systems:

• Intel Mac

• Solaris 64-bit

Support has been dropped for the Solaris 2 operating system. It has been replaced by support for Solaris 64.

A readme file containing a customized checklist of deployment prerequisites is now generated in the output folder with each compiled application.

While still supported in this release, customers should prepare for discontinuance of support for the following Borland compilers:

• Borland C++Builder version 6.0

• Borland C++Builder version 5.0

• Borland C/C++ (free command-line tools) version 5.5

MATLAB Compiler4.6 includes the-Foption for bothmccanddeploytool. Use this option to feed back a specific project file's settings toMATLAB Compiler. For more information, see theMATLAB Compiler User's Guidedocumentation.

以下是与Windows Vista™已知问题as of this release. To resolve these issues, ensure you are logged in as Administrator.

• 用户帐户控制(UAC)启用,standard user is not able to write to a folder in thec:\Program Filesfolder. See theMATLAB Release Notesfor more details regarding this issue.

• 用户帐户控制(UAC)启用,standard user is not able to register DLLs. Whenmbuild -setupattempts to registermwcomutil.dll, the following errors are displayed:

  Trying to update options file: C:\Users\qe\AppData\Roaming\MathWorks\MATLAB\R2007a\compopts.bat From template: C:\PROGRA~1\MATLAB\R2007a\bin\win64\mbuildopts\msvc80compp.bat Done . . . --> "C:\PROGRA~1\MATLAB\R2007a\bin\win64\mwregsvr C:\PROGRA~1\MATLAB\R2007a\bin\ win64\mwcomutil.dll" Error: DllRegisterServer in C:\PROGRA~1\MATLAB\R2007a\bin\win64\mwcomutil.dll failed Undefined subroutine &mexsetup::expire called at C:\PROGRA~1\MATLAB\R2007a\bin\/ mexsetup.pm line 839.

  This is also the case when MATLAB Builder for Excel invokesmbuildin an attempt to register DLLs after a compilation completes.

• The following message is displayed when MATLAB Builder for .NET, attempts to install a DLL into the Global Assembly Cache from a network location:

  Unhandled Exception: System.IO.FileLoadException: could not load file or assembly 'GACInstaller, Version=1.0.2568.30711, Culture=neutral, PublicKeyToken=null' or one of its dependencies. Failed to grant permission to execute. (Exception from HRESULT: 0x80131418) File name: 'GACInstaller, Version=1.0.2568.30711, Culture=neutral, PublicKeyToken=null' ---> System.Security.Policy.PolicyException: Execution permission cannot be acquired. at System.Security.SecurityManager.ResolvePolicy(Evidence evidence, PermissionSet reqdPset, PermissionSet optPset, PermissionSet denyPset, PermissionSet& denied, Boolean checkExecutionPermission) at System.Security.SecurityManager.ResolvePolicy(Evidence evidence, PermissionSet reqdPset, PermissionSet optPset, PermissionSet denyPset, PermissionSet& denied, Int32& securitySpecialFlags, Boolean checkExecutionPermission)

• Printing from a compiled application is not currently available with Microsoft Windows Vista. The following message is displayed if you attempt to print on systems with the Microsoft Windows Vista 32-bit operating system installed:

  PrintImage Error StartPage failed with error 6: The handle is invalid.

  Printing with Vista 64-bit installed results in an error message that lists a manifest as missing. Please check//www.tianjin-qmedu.com/support/for updates on these issues.

MATLAB Compiler, in conjunction with the optionalMATLAB Builder™JA, lets you convert MATLAB files into Java components.

As an alternative to themcccommand to invokeMATLAB Compiler, you can invoke the graphical user interface forMATLAB Compilerby issuing the following command at the MATLAB prompt:

Use the Deployment Tool to perform the tasks shown in the following conceptual illustration:

This release provides support for a new target, .NET components. With the optional MATLAB Builder for .NET product, you can create both COM components that can be used in native code applications and .NET components that can be used in managed code applications.

MATLAB Compiler, in conjunction with the optional MATLAB Builder for .NET, lets you convert MATLAB files into .NET components that are accessible from any Common Language Specification (CLS)-compliant client code. In addition, you can build Common Object Model (COM) components that are accessible from Visual Basic, C/C++, Microsoft Excel, or any other COM client.

MATLAB Compiler supportsMicrosoft Visual C++Version 8.0 as a third-party compiler. However, if you useMicrosoft Visual C++Version 8.0 to generate applications/components, you must have theMicrosoft Visual Studio2005 run-time libraries available on the computer used for deployment. If you distribute your applications/components, you must make sure that any machine used for deployment has these run-time libraries installed. For more information, seesolution 1–2223MW.

MATLAB Compiler4.4 (R2006a) does not support HP-UX.MATLAB Compiler4.3 (R14SP3) was the last release to support HP-UX.

You can useMATLAB Compiler4.3 on the following supported systems to create redistributable, standalone applications or software components. These applications or components can then be deployed to other systems with the same operating system.

Supported Platforms

MATLAB Compiler4.3 is supported on these platforms:

• Windows

• Linux

• Solaris™

• HP-UX^®

• Linux x86-64

• Mac OS^®X

TheMATLAB Compilerdocumentation has been updated to include configuration information for all the supported platforms.

You can useMATLAB Compiler4.2 on the following supported systems to create redistributable, standalone applications or software components. These applications or components can then be deployed to other systems with the same operating system.

Supported Platforms

MATLAB Compiler4.2 is supported on these platforms:

• Windows

• Linux

• Solaris

• HP-UX

• Linux x86-64

TheMATLAB Compilerdocumentation has been updated to include configuration information for all the supported platforms.