How does one use Obj-C 2.0 with GNUstep? - objective-c

I'm aware of the existence of libobjc2, and I gather that I'll need to use clang rather than GCC, but I can't find any basic instructions of what's different about the compilation process.
Can anyone give explicit, step-by-step instructions on acquiring, configuring, compiling, and using GNUstep with Objective-C 2.0 in Ubuntu?

Since the GNUStep ObjC2 FAQ has already been referenced, I'll assume you've had a look. I would like to point this out, however:
For more advanced features, currently only supported if you compile with Clang, you will need the GNUstep runtime. This is not currently considered production ready. It should work as a drop-in replacement for the GCC runtime, but some of the advanced features are not well-tested.
The FAQ also calls out the following:
If you are using a recent version of GCC to compile your code then you should have a copy of the Objective-C runtime library that came with your compiler.
If you compile with the -fobjc-nonfragile-abi flag then you will use the new ABI. This is only supported with the GNUstep runtime. This adds:
Property introspection
Introspection on optional protocol methods
Non-fragile instance variables
Forwarding proxy support
So, if you're looking for older functionality via GCC, this tutorial would do the job nicely, as would this one. You've explicitly called out that you're looking to do this with the functionality provided by Clang (via libobjc2), which has its own bed of documentation.
Obtaining a release of Clang and getting set up seems fairly cut and dried. The LLVM Getting Started document is exceptionally robust (full documentation archive here), which should provide insight into getting the backend set up and compiling properly. Some additional insight into the expected build procedure can be found here.
However, compiling an Objective-C application in Clang seems extremely underdocumented and untested at the time of this writing. As features become available and the codebase stabilizes, I imagine more user documentation will go live.

According to the GNUstep Wiki, you need just need to compile your code with -fblocks and -fobjc-nonfragile-abi. Then you just use the language features in your code.

Related

Mono to Emscripten

Mono has a LLVM compiler. Is there anyway to use it with Emscripten (compile C# code to JavaScript)?
There is currently no out of the box way to do this. It might be possible, but it would require a lot of work. You would need to run mono in full AOT (ahead of time) compilation mode with the LLVM codegen. But there are many issues:
LLVM is currently not used for all methods, and mono falls back to it's own code generator in a number of cases. You would either need to get the LLVM suport working for all cases, or provide the JS code needed when LLVM cannot be used.
Mono currently has a number of architecture specific files (x86, amd64, arm, etc) and would probably need equivalent for JS, both for the code generation and for the AOT runtime.
And so on...
you can try to use C# Native
have a look here http://csnative.codeplex.com
Even if you run mono in full AOT and compile your program with LLVM it is not possible to use it with emscripten. This is quote from my discussion in mono group:
Besides that, no, it's not possible to use emscripten with mono's llvm output for a lot of reasons.
The output embeds calls to the mono runtime and some trampilines.
Mono use a custom LLVM with custom IR operations and that won't work on emscripten without some work on their end.

Linux Clang and Objective-C base library

I have been experimenting with Objective-C using GCC + GNUstep on an Ubuntu system.
Now regarding the LLVM Clang compiler, what kind of *step library does it offer? Does it use the GNUstep on the Apple Cocoa? I am mostly interested in the base library - collections, streams, etc. The website doesn't give much information.
Like MKroehnert and puzzle said, neither LLVM Clang nor GCC actually come with a set of frameworks. GCC comes with only a small runtime that doesn't provide NSArray, NSString, not even NSObject.
Frameworks are provided by, for example:
GNUstep
Cocotron
Cocoa
ObjFW
ObjFW is the lightest of them all, but on any UNIX, I'd recommend you go the GNUstep route.
LLVM Clang is a compiler. It is completely independent from the Cocoa / Cocoa Touch frameworks on OS X / iOS, or any other frameworks or platform, for that matter.
Like puzzle said in his answer clang is a compiler like gcc.
On linux you can also use clang instead of gcc together with the GNUstep libraries (which provide the NS* classes you were asking about).
For more information see GNUstep Objc2 FAQ or this blogpost for example.
Like others said, LLVM Clang is just a compiler. But you can have all the modern features of Objective-C 2.0 on Ubuntu using Clang, the modern GNUstep Objective-C 2.0 runtime and GNUstep itself. Have a look here for a how to:
http://wiki.gnustep.org/index.php/GNUstep_under_Ubuntu_Linux
That page is maintained by the GNUstep developers and contains bash scripts to compile and install everything needed for Objective-C 2.0 from scratch for different versions of Ubuntu-Linux using Clang and the GNUstep Objective-C runtime, which can be found here: https://github.com/gnustep/libobjc2 . It would be moot to copy those lengthy scripts to Stackoverflow since they would get out of date sooner or later. So follow that link to get everything from first hand.

What is the difference between the different compilers

In the cocoa fundamentals documentary, it mentions about the different compilers. The gcc, LLVM-GCC and Clang. Does anyone specifically know what they do and when to use them?
They all do the same thing, produce executable object code that can be linked together to build a program (or shared library). They just do it in different ways.
gcc is a venerable old compiler of yore that is getting a bit long in the tooth, but is still the primary workhorse of virtually the entire open source world. It is also the most portable compiler in existence.
LLVM-GCC is a component of the LLVM project that uses the gcc parser front end with the LLVM code-generation backend.
Clang is an LLVM-specific front end that uses the LLVM back end. It is essentially a clean-room compiler that promises faster compiles producing more efficient code than gcc, all from a much smaller and cleaner code base (thanks #dreamlax).
Much more information on this is available on the LLVM site.
gcc: GNU Compiler Collection. Very portable compilers for many different languages. Licensed under the GPL.
LLVM-gcc: Uses the frontend of GCC to parse the program code but uses LLVM to generate machine code.
Clang: New frontend for C-Like languages (C, C++, Objective-C) that uses LLVM to generate machine code.
Apple recommends you always use Clang. The main advantages of Clang are that it is faster and that it generates better diagnostic messages (warnings and errors). Also the generated code from clang is often faster than the one from GCC.
The version of Clang that is shipped with Xcode 3 doesn’t yet support C++ (or Objective-C++), so Xcode will use LLVM-GCC for C++.
Here is a more detailed comparison of Clang and GCC.

how to build g++

I'm currently trying to get g++ working, and looking at http://gcc.gnu.org/install/build.html, I can't seem to find where it says how "to perform a 3-stage bootstrap of the compiler". Where would I find this information?
(I'm on a mac, in case that matters.)
After running configure, you do make bootstrap
You cannot bootstrap GCC without an already-functioning compiler on your platform:
From the prerequisites page:
ISO C90 compiler
Necessary to bootstrap GCC, although versions of GCC prior to 3.4 also allow bootstrapping with a traditional (K&R) C compiler.
And to preempt your next question, you also need a functioning compiler (usually GCC) in order to build Clang+LLVM.
It might be possible to do what you're trying so hard to do, by cross-compiling GCC from some other platform (see this section of the documentation). But I imagine this is not for the feint of heart, and you'd be far, far better off simply updating your copy of Mac OS X to the latest version with Xcode support.
From that link:
For a native build, the default configuration is to perform a 3-stage bootstrap of the compiler when `make' is invoked.
It seems to me that if your configuration isn't tweaked, it should do it out of the box. Just type make.
More specifically, you have to download the source code, and follow the instructions in that whole tutorial in order to build.
A side note - I am finding it hard to believe that there is no easy way to get GCC on an OSX box without having the installation media. That sounds really annoying :)
Edit:
If you are simply trying to write C++ on OSX, you could install one of many other IDEs. If you are lucky, they may come with their own compiler. Here is a list of alternatives to XCode:
http://alternativeto.net/desktop/xcode/?profile=mac&platform=mac

Objective-C for Windows

What would be the best way to write Objective-C on the Windows platform?
Cygwin and gcc? Is there a way I can somehow integrate this into Visual Studio?
Along those lines - are there any suggestions as to how to link in and use the Windows SDK for something like this. Its a different beast but I know I can write assembly and link in the Windows DLLs giving me accessibility to those calls but I don't know how to do this without googling and getting piecemeal directions.
Is anyone aware of a good online or book resource to do or explain these kinds of things?
Expanding on the two previous answers, if you just want Objective-C but not any of the Cocoa frameworks, then gcc will work on any platform. You can use it through Cygwin or get MinGW. However, if you want the Cocoa frameworks, or at least a reasonable subset of them, then GNUStep and Cocotron are your best bets.
Cocotron implements a lot of stuff that GNUStep does not, such as CoreGraphics and CoreData, though I can't vouch for how complete their implementation is on a specific framework. Their aim is to keep Cocotron up to date with the latest version of OS X so that any viable OS X program can run on Windows. Because GNUStep typically uses the latest version of gcc, they also add in support for Objective-C++ and a lot of the Objective-C 2.0 features.
I haven't tested those features with GNUStep, but if you use a sufficiently new version of gcc, you might be able to use them. I was not able to use Objective-C++ with GNUStep a few years ago. However, GNUStep does compile from just about any platform. Cocotron is a very mac-centric project. Although it is probably possible to compile it on other platforms, it comes XCode project files, not makefiles, so you can only compile its frameworks out of the box on OS X. It also comes with instructions on compiling Windows apps on XCode, but not any other platform. Basically, it's probably possible to set up a Windows development environment for Cocotron, but it's not as easy as setting one up for GNUStep, and you'll be on your own, so GNUStep is definitely the way to go if you're developing on Windows as opposed to just for Windows.
For what it's worth, Cocotron is licensed under the MIT license, and GNUStep is licensed under the LGPL.
You can use Objective C inside the Windows environment. If you follow these steps, it should be working just fine:
Visit the GNUstep website and download GNUstep MSYS Subsystem (MSYS for GNUstep), GNUstep Core (Libraries for GNUstep), and GNUstep Devel
After downloading these files, install in that order, or you will have problems with configuration
Navigate to C:\GNUstep\GNUstep\System\Library\Headers\Foundation1 and ensure that Foundation.h exists
Open up a command prompt and run gcc -v to check that GNUstep MSYS is correctly installed (if you get a file not found error, ensure that the bin folder of GNUstep MSYS is in your PATH)
Use this simple "Hello World" program to test GNUstep's functionality:
#include <Foundation/Foundation.h>
int main(void)
{
NSAutoreleasePool * pool = [[NSAutoreleasePool alloc] init];
NSLog(#"Hello World!.");
[pool drain];
return;
}
Go back to the command prompt and cd to where you saved the "Hello World" program and then compile it:2
gcc -o helloworld.exe <HELLOWORLD>.m -I /GNUstep/GNUstep/System/Library/Headers -L /GNUstep/GNUstep/System/Library/Libraries -std=c99 -lobjc -lgnustep-base -fconstant-string-class=NSConstantString
Finally, from the command prompt, type helloworld to run it
All the best, and have fun with Objective-C!
NOTES:
I used the default install path - adjust your command line accordingly
Ensure the folder path of yours is similar to mine, otherwise you will get an error
Also:
The Cocotron is an open source project which aims to implement a cross-platform Objective-C API similar to that described by Apple Inc.'s Cocoa documentation. This includes the AppKit, Foundation, Objective-C runtime and support APIs such as CoreGraphics and CoreFoundation.
http://www.cocotron.org/
WinObjC? Windows Bridge for iOS (previously known as ‘Project Islandwood’).
Windows Bridge for iOS (also referred to as WinObjC) is a Microsoft open source project that provides an Objective-C development environment for Visual Studio/Windows. In addition, WinObjC provides support for iOS API compatibility. While the final release will happen later this fall (allowing the bridge to take advantage of new tooling capabilities that will ship with the upcoming Visual Studio 2015 Update),
The bridge is available to the open-source community now in its current state. Between now and the fall. The iOS bridge as an open-source project under the MIT license. Given the ambition of the project, making it easy for iOS developers to build and run apps on Windows.
Salmaan Ahmed has an in-depth post on the Windows Bridge for iOS http://blogs.windows.com/buildingapps/2015/08/06/windows-bridge-for-ios-lets-open-this-up/ discussing the compiler, runtime, IDE integration, and what the bridge is and isn’t. Best of all, the source code for the iOS bridge is live on GitHub right now.
The iOS bridge supports both Windows 8.1 and Windows 10 apps built for x86 and x64 processor architectures, and soon we will add compiler optimizations and support for ARM, which adds mobile support.
I have mixed feelings about the Cocotron project. I'm glad they are releasing source code and sharing but I don't feel that they are doing things the easiest way.
Examples.
Apple has released the source code to the objective-c runtime, which includes properties and garbage collection. The Cocotron project however has their own implementation of the objective-c runtime. Why bother to duplicate the effort? There is even a Visual Studio Project file that can be used to build an objc.dll file. Or if you're really lazy, you can just copy the DLL file from an installation of Safari on Windows.
They also did not bother to leverage CoreFoundation, which is also open sourced by Apple. I posted a question about this but did not receive an answer.
I think the current best solution is to take source code from multiple sources (Apple, CocoTron, GnuStep) and merge it together to what you need. You'll have to read a lot of source but it will be worth the end result.
I'm aware this is a very old post, but I have found a solution which has only become available more recently AND enables nearly all Objective-C 2.0 features on the Windows platform.
With the advent of gcc 4.6, support for Objective-C 2.0 language features (blocks, dot syntax, synthesised properties, etc) was added to the Objective-C compiler (see the release notes for full details). Their runtime has also been updated to work almost identically to Apple's own Objective-C 2.0 runtime. In short this means that (almost) any program that will legitimately compile with Clang on a Mac will also compile with gcc 4.6 without modification.
As a side-note, one feature that is not available is dictionary/array/etc literals as they are all hard-coded into Clang to use Apple's NSDictionary, NSArray, NSNumber, etc classes.
However, if you are happy to live without Apple's extensive frameworks, you can.
As noted in other answers, GNUStep and the Cocotron provide modified versions of Apple's class libraries, or you can write your own (my preferred option).
MinGW is one way to get GCC 4.6 on the Windows platform, and can be downloaded from The MinGW website. Make sure when you install it you include the installation of C, C++, Objective-C and Objective-C++. While optional, I would also suggest installing the MSYS environment.
Once installed, Objective-C 2.0 source can be compiled with:
gcc MyFile.m -lobjc -std=c99 -fobjc-exceptions -fconstant-string-class=clsname (etc, additional flags, see documentation)
MinGW also includes support for compiling native GUI Windows applications with the -mwindows flag. For example:
g++ -mwindows MyFile.cpp
I have not attempted it yet, but I imagine if you wrap your Objective-C classes in Objective-C++ at the highest possible layer, you should be able to successfully intertwine native Windows GUI C++ and Objective-C all in the one Windows Application.
Check out WinObjC:
https://github.com/Microsoft/WinObjC
It's an official, open-source project by Microsoft that integrates with Visual Studio + Windows.
If you just want to experiment, there's an Objective-C compiler for .NET (Windows) here: qckapp
You can get an objective c compiler that will work with Windows and play nice with Visual Studio 2008\2010 here.
open-c flite
Just download the latest source. You don't need to build all of CF-Lite there is a solution called objc.sln. You will need to fix a few of the include paths but then it will build just fine. There is even a test project included so you can see some objective-c .m files being compiled and working in visual studio. One sad thing is it only works with Win32 not x64. There is some assembly code that would need to be written for x64 for it to support that.
A recent attempt to port Objective C 2.0 to Windows is the Subjective project.
From the Readme:
Subjective is an attempt to bring Objective C 2.0 with ARC support to
Windows.
This project is a fork of objc4-532.2, the Objective C runtime that
ships with OS X 10.8.5. The port can be cross-compiled on OS X using
llvm-clang combined with the MinGW linker.
There are certain limitations many of which are a matter of extra
work, while others, such as exceptions and blocks, depend on more
serious work in 3rd party projects. The limitations are:
• 32-bit only - 64-bit is underway
• Static linking only - dynamic linking is underway
• No closures/blocks - until libdispatch supports them on Windows
• No exceptions - until clang supports them on Windows
• No old style GC - until someone cares...
• Internals: no vtables, no gdb support, just plain malloc, no
preoptimizations - some of these things will be available under the
64-bit build.
• Currently a patched clang compiler is required; the patch adds
-fobjc-runtime=subj flag
The project is available on Github, and there is also a thread on the Cocotron Group outlining some of the progress and issues encountered.
Get GNUStep here
Get MINGW here
Install MINGW
Install GNUStep
Then Test
If you are comfortable with Visual Studio environment,
Small project: jGRASP with gcc
Large project: Cocotron
I heard there are emulators, but I could find only Apple II Emulator http://virtualapple.org/. It looks like limited to games.
First of all, forget about GNUStep tools. Neither ProjectManager nor ProjectCenter can be called an IDE. With all due respect, it looks like guys from GNUStep project are stuck in the late 80-s (which is when NeXTSTEP first appeared).
Vim
ctags support Objective-C since r771 (be sure to pick the pre-release 5.9 version and add --langmap=ObjectiveC:.m.h to the command line, see here), so you'll have decent code completion/tag navigation.
Here's a short howto on adding Objective-C support to Vim tagbar plugin.
Emacs
The same applies to etags shipped with modern Emacsen, so you can start with Emacs Objective C Mode. YASnippet will provide useful templates:
and if you want something more intelligent than the basic tags-based code completion, take a look at this question.
Eclipse
CDT supports Makefile-based projects:
-- so technically you can build your Objective-C projects out of the box (on Windows, you'll need the Cygwin or MinGW toolchain). The only problem is the code editor which will report plenty of errors against what it thinks is a pure C code (on-the-fly code checking can be turned off, but still...). If you want proper syntax highlighting, you can add Eclim to your Eclipse and enjoy all the good features of both Eclipse and Vim (see above).
Another promising Eclipse plugin is Colorer, but it doesn't support Objective-C as of yet. Feel free to file a feature request though.
SlickEdit
SlickEdit, among other features of a great IDE, does support Objective-C. While it is fairly complex to learn (not as complex as Emacs though), I believe this is your best option provided you don't mind purchasing it (the price is quite affordable).
Additionally, it has an Eclipse plugin which can be used as an alternative to the stand-alone editor.
KDevelop
Rumor has it there exists a KDevelop patch (15 year old, but who cares?). I personally don't think KDevelop is feature-superior compared to Emacsen, so I wouldn't bother trying it.
The above also applies to Objective-C development on Linux, since all of the tools mentioned are more or less portable.
As of 2021, the GNUstep Windows MSVC Toolchain allows to integrate Objective-C code in any Windows app, including Visual Studio projects using LLVM/Clang. This includes support for Automatic Reference Counting (ARC) and Objective-C 2.0 features such as blocks.
The project includes the Foundation, CoreFoundation, and libdispatch libraries from GNUstep. It does currently not include any UI framework such as AppKit or UIKit, but it can be used to e.g. write a Windows-specific UI with cross-platform business logic written in Objective-C.