Installing g++ to terminal from files on my mac - c++

When I try to compile a c++ program in my Mac terminal, I get the following error:
-bash: g++: command not found
but I believe I have all the files I need on my Mac to run the compiler. I have the Developer folder, which contains Xcode in Applications, g++ in the usr/bin folder, and a bunch of other folders.
How can I use these files to install the g++ compiler. Thank you!

Assuming you have the latest Xcode, go to Xcode Preferences, Downloads, and click on the Install button next to Command Line Tools. This should install and put everything into your PATH automatically.

houbysoft's answer above is correct, but the OP needs a little more detail.
The basic idea is that you need the "Command Line Tools" environment installed. But the name for that environment, and how you install it, has changed multiple times over the years. So, houbysoft's steps work for 4.3.3, but not for 3.2.6.
In the 3.2 era, the tools were called "UNIX Development". And, rather than being downloaded and installed from within Xcode, they came as part of the Xcode mpkg and were installed as part of the initial Xcode install. In most versions, there was a checkbox named "UNIX Development", usually checked by default, but in some versions it was a separate step. Looking at the "Xcode 3.2.6 and IOS SDK 4.3" disk image currently available from Apple, it's a checkbox. Anyway, that's the only officially-supported way to get them, but lots of other ways work… Here are your options:
Throw away Xcode 3.2.6 (just trash the whole /Developer) and install 4.3.3, then follow houbysoft's steps to get the Command Line Tools. Unless there's a good reason you can't use 4.3, this is almost certainly the best answer.
Throw away Xcode 3.2.6 and reinstall it, and this time make sure the "UNIX Development" checkbox is on.
Just re-run the 3.2.6 installer, and make sure the "UNIX Development" checkbox is on.
Log into developer.apple.com and look through the downloads for a package named "UNIX Development Tools", "Developer Tools CLI", "CLI Developer Environment", "Command Line Tools", or similar that corresponds to 3.2.6. There is such a package for most, but not all, versions of Xcode, and can be used to set up a command-line build environment either with or without Xcode.
Look at the invisible packages on the Xcode disk image. I don't remember whether you need just DeveloperToolsCLI.pkg, or that plus a few others, but a bit of trial and error or googling should get you there.

If everything is installed correctly, it sounds like all you need to do is get your "PATH" environment variable modified.
This closely related SuperUser question should provide the answers you need, at least if you're running under Snow Leopard. Here's another question about setting environment variables for Lion.

Related

Fixing MinGW Installation on Windows 8

While helping my friend spin up MinGW and a C++ environment on his Windows 8 computer, I ran the get-mingw script and waited as it ran through all the mirrors for required downloads. However, three downloads completely failed:
libltdl - installer script hung and then moved on after pressing "OK"
automake-1.11 - installer script tried finding 1.10, then 1.9, then 1.8, then 1.7 (all of which failed) until finally settling on 1.6
mktemp - script hung and moved on after pressing "OK"
In all three cases, the script gave me a nice error log upon completion, showing that a majority of packages had been downloaded and installed except for these three, which showed up as errors. However during the installation process I had simply gone to the MinGW sourceforge page and manually found and downloaded each .bin.tar.lzma file that was missing.
Now that I have them, is there a good accepted way to unpack and plug them into my friend's existing MinGW install? In case it's tough, I'm comfortable with unix and dos command line so I'll be able to move executables into the MinGW/bin folder if that's what's needed, I just want to check for the best way to 'fix' the install.
As a side note - even though the error log says these are required packages, adding MinGW/bin/ to the PATH still allows for use of gcc and g++, although not make (possibly because of automake failure?). Is this standard behavior?
Firstly, the package issue can be fixed by using the MinGW installer - keep the packages selected and go to "apply changes" and the script will probably try to redownload the missing packages. I think the original problem was probably just a shoddy wifi connection during repository connection.
However, I then ran into a problem where I tried to run gcc and it gave me a missing -lpthread error ... but this question was able to help me fix that, and gcc and g++ are working fine now (haven't opened and tested Eclipse yet though). Just in case of link decay, the issue I cited arises from the MinGW installer script not downloading the lpthread library upon installation. To fix that issue, quoted from link:
Just run and open MinGW Installation Manager, which should be pre-installed with MinGW, select "All Packages" on the left panel, and on the right panel, search for "mingw32-pthreads-w32" packages and install them.
I think the Installation Manager has libpthread and pthread available for install, and pthread libs were the ones that seemed to solve it for me.

Building C++ not working in OSX 10.9

Update:
I solved the problem in the end by reinstalling command line tools, as shown in this link.
Sorry for the trouble!
Initial question
I have just upgraded to OSX Mavericks and I cannot make the compiler work anymore.
I've been using gcc48 from macports with vim/sb2 and everything was working just fine. Now any library that I include gives tons of errors. I am not very experienced and I do not know what to do, or what to search for. I just want it to work like before. Here is what is says when compiling a basic "hello world" program:
In file included from /opt/local/include/gcc48/c++/bits/postypes.h:40:0,
from /opt/local/include/gcc48/c++/iosfwd:40,
from /opt/local/include/gcc48/c++/ios:38,
from /opt/local/include/gcc48/c++/ostream:38,
from /opt/local/include/gcc48/c++/iostream:39,
from ceva.cpp:1:
/opt/local/include/gcc48/c++/cwchar:44:19: fatal error: wchar.h: No such file or directory
#include <wchar.h>
If you need any information, tell me and I will provide it to you. Thank you for your patience!
[Added 2019 - Mojave/Catalina] XCode 10 has removed the placement of headers in /usr/include, even with the command line tools. If you want headers, and you're on you should be able to install a transitional package, which is available on Mojave. This package is not present on the Catalina. The headers are within the SDK for the appropriate release.
XCode 5/Mavericks have changed the defaults for where headers are located.
On prior versions of Mac OS X / XCode you would have found headers in /usr/include.
When you built gcc on the older release it picked up the headers from that location, and looks for them there now when you try to build code.
The problem is that the files are not there; they're somewhere under /Applications/Xcode.app/Contents/Developer.
You could futz about with a spec file to get it working consistently again, but the sanest thing to do is rebuild gcc.
As has been mentioned as well, you can install the command line tools using xcode-select --install, which will reinstall the developer tools, including the headers in /usr/include.
I tend to reinstall things like gcc when I upgrade my operating system as I encounter other errors due to changes in the environment. This is just a habit I've formed. It may not be the best habit, but it's saved me pain on numerous occasions.
[added 2017] As an addendum, if you install the command line tools, then the headers will be placed in /usr/include as well. The command to install the command line tools is xcode-select --install - this will allow you to use most compilers without having to specify the location of the headers manually.
This is how I got it working on Mac osx Mavericks:
Open terminal
type --> xcode-select --install
A pop-up windows will apear asking you about install tools
choose install tools
wait install to finish
Here is a manual method for updating XCode on OS X Mavericks:
Go into the developers site http://developer.apple.com
Then go to Downloads for Apple developers and there is a command line tools installer for Mavericks — here's the link (but you'll need a developers account for the link to work) http://developer.apple.com/downloads/index.action?q=xcode
I'm using MacOs Sierra 10.12.4, got the same problem.
I solved it by using:
/usr/bin/gcc
For me the fix was to uninstall old broken gcc:
brew uninstall gcc#5
I noticed paths like /usr/local/Cellar/gcc#5 in build output, while gcc#8 was also installed and is the one that was supposed to be used.

Getting mingw-get to install correctly - mingw/msys path missing plus more!

I'm running windows XP.
I have been following this tutorial and so downloading mingw-get-inst here.
I've done this a couple times and the last time I checked boes to install EVERYTHING including, but not limited to, gcc, g++, MSYS and the MinGW Compiling Suite. I told it to make a program menu shortcut, too.
I believe I have correctly added MinGW/bin to the Environment Settings' Path.
So I have the shortcut to the MSYS shell now. It goes to C:\MinGW\msys\1.0\msys.bat, but the whole msys directory does not exist!
There are tons of "packages" have been downloaded as far as I can tell ("mingw-get show" brings up a lot!), but it doesn't seem like any of it installed correctly.
Also, should I be able to use commands like "g++ foo.cpp -o foo.exe" from cmd.exe or only from the MSYS shell? Because I was unable to use cmd.exe.
Please help!
The MSYS and MinGW installation procedure is a nightmare, and has got worse over the years. If you just want the compiler suite (no UNIX tools like sed, grep), then I suggest you go to Twilight Dragon Media and download GCC from there. If you also want all the tools, including the Bash shell, the easiest way I've found to get them is to install Git for Windows (which you may want anyway), which includes them all. And if you want a C++ IDE that works well with GCC, take a look at Code::Blocks.
What is the result of running, at a cmd.exe prompt, the command echo %Path%? It should (given what you wrote about your system) include “C:\MinGW\bin”—does it?
What is the result of running dir C:\MinGW?
Note that mingw-get show shows details of available packages—those mingw-get knows how to download and install. The installer is a bit primitive, but has gotten better over the last year or so.
If you issue the command $ echo $PATH, you'll notice it contains /mingw/bin But the /mingw directory is unavailable. To mount it, edit the MSYS fstab, adding a line like C:\MinGW /mingw For details, see the MingW Getting Started guide, more specifically section "After Installing You Should..."
You only need to add MinGW/bin to the Windows' Environment Settings' Path if you plan to use MingW's executables outside of MSYS.

The equivalent of ./configure in Windows?

What is the equivalent of ./configure in Windows?
Sometimes I download a C/C++ library and when I use the make it, it says "use ./configure" but obviously ./configure can only be used on a Linux machine and the libraries don't usually have instructions for compiling on Windows (although they do support Windows, they don't provide instructions).
For example, the library wxSVG says it works on Windows, but when I download it I don't see any instructions for compiling on Windows, and I only Linux files for configuring it.
I just faced with the same issue and here is what I did,
I first installed MinGw using the installation manager (with msys base included). Then I go to C:\MinGW\msys\1.0\ folder in my pc, where msys.bat (to evoke the MinGW shell) and run it. Then on that bash screen, I navigated to the folder that I wanted to install initially. After that, using "./configure" and "make" worked just fine.
Actually, ./configure is not Linux-specific at all. Its original purpose was to smooth over the differences between the many variants of Unix now thankfully relegated to the dust heap of history, but nowadays it may well know how to set up things to work on Windows.
I would install the MinGW/MSYS development tools and see if the configure script is happy in that environment. (If that doesn't work, I can't help you any further.)
./configure is a script that comes with the source you have downloaded. You will use it the same on windows as you do on any other operating system. Unfortunately, you will need a posix-like shell to run it. A good option for that is to use mingw or cygwin
I'd say that this is program dependant. You see "configure" is a program/script in the local directory, it's not a global command/program (like "dir" would be)...
You either have luck to find something like "configure.bat" or "configure.cmd", or you'd have to adapt the configure-file into a BATCH-file.
You only run the ./configure command when building certain applications from source.
So Unzip it where you want to install it and then go to the folder where you unziped it and run "./configure"

Eclipse Ganymede and MinGW in Windows

I'm trying to get eclipse to work with MinGW.
I've done the following:
Downloaded CDT for eclipse.
Installed MinGW.
Added C:\MinGW\bin to my path.
Opening a command prompt (CMD) and typing g++ or alike works fine.
I run eclipse, create a "New C++ Project", and only get the option saying "other toolchains".
There's a MILLION tutorials out there saying eclipse should identify MinGW on its own. It doesn't, and I don't know what to do. I've tried reinstalling everying in just about every order posible. Still no luck.
I've also noted some tutorials say something about creating a "Managed C++ Project". I've no such option, all I get is "C++ Project" and "C Project"
edit:
I have eclipse ganymede, windows x86_64, version 3.4.2
http://download.eclipse.org/eclipse/downloads/drops/R-3.4.2-200902111700/index.php
Running the "Eclipse IDE for C/C++ developers" fails, since there's no x64 version for windows. The x86 version requires x86 JAVA installed as well, and installing two versions of java, gave nothing but trouble in the past.
The instructions for setting up MinGW in Ganymede are located here.
The following are instructions and
links on how to install the current
version of MinGW. Note that these
links may become inaccurate over time
as new versions of MinGW components
are introduced. Please check the MinGW
File Release section for the latest
versions.
Download and run the MinGW setup program, MinGW-5.1.3.exe.
Select download and install the MinGW base tools and the g++ compiler.
You may select the Current or
Candidate version of these tools. You
may also install any of the other
available compilers as well.
Do not install the MinGW Make feature as the MSYS version of make
from step 5 is a more complete
implementation of make.
The MinGW setup program currently does not install the gdb
debugger. To install the debugger,
download the file from the following
location: gdb-6.6.tar.bz2
Extract the contents of the file gdb-6.6.tar.bz2 to the same location
where you installed MinGW.
If you want to use Makefile projects, download and run the setup
program from the following location:
MSYS-1.0.10.exe. MSYS provides an
implementation of make and related
command line tools. This is not
required for other types of projects
with the MinGW toolchain, which use
CDT's internal build tools to perform
the build.
Following this process resolved any problems I had.
I had the same exact problem with Eclipse Galileo and CDT 6.0.1. It turns out that CDT only recognized MinGW when it's located under c:\mingw. I had it in c:\msys\mingw so that was the problem. After I changed that everything worked fine.
The distinction between managed make projects and makefile project was removed in CDT 4.x, I think. Now there is only one type of project, but you can select different builders. CDT includes an internal builder which does not use makefiles and another one which does.
First, save yourself the effort of "reinstalling in every order possible". That is also known as trial-and-error, and will only make you more frustrated. Apply the normal problem-solving skills you have as a programmer.
Given that you have MinGW installed, what happens if you download "Eclipse IDE for C/C++ developers", start eclipse.exe, and try to create a C++-project with a MinGW toolchain?
EDIT: remember: the key in getting help with problems like these is to produce a minimal example which fails. Also, it would help if you provided URLs to the packages you installed (MinGW, Eclipse, etc.).
EDIT: I just installed CDT using the Ganymede update site, downloaded and installed MinGW from here, and restarted Eclipse, and everything worked fine. I know that doesn't help you, but it does prove that the toolchain detection isn't completely broken. Something is weird on your side.
You could try Wascana Desktop Developer. Its a distribution of Eclipse CDT configured specifically for developing on Windows.
I had the same problem (i.e. Eclipse not finding MinGW on the PATH) after I removed some of the unused files/folders from MinGW. It was ~600 MB and I was tasked to trim it down before adding to source control. I got it down to a workable ~200 MB. When I tried to re-create an Eclipse workspace afterwards, MinGW disappeared from available toolchains. It reappeared after I put the original MinGW install on the path.
HTH