How to create symbolic links in the output .tar.gz file? - ant

We are using Windows as a development system and Ant to create the platform-specific bundles. For the macOS specific bundle (.tar.gz file) we are using the tar task. I want to create a symbolic link in the output .tar.gz file which points to another file in the same .tar.gz file. Can this be done using Ant?

To the best of my knowledge stock Ant can't do this.

Bit late, but suggestion: you can create tar archives containing symlinks from windows command line using cygwin.
$ ln -s abc def
$ tar cf test.tar abc def
$ tar tvf test.tar
-rw-r--r-- shegny/Domain Users 0 2013-04-26 10:25 abc
lrwxrwxrwx shegny/Domain Users 0 2013-04-26 10:25 def -> abc

You cannot create a symbolic link on Windows. Therefore, the Symlink Ant task would not work properly. One option would be to run a continuous build server on a Mac machine. Then, you could run that task at regular intervals and the Symlink Ant task would work.

Related

How to build nix hello world

I'm trying to get started with a simple Hello World derivation using the Nix manual.
But it's not clear to me how to go about building it.
Is there somewhere I can download the source files from so I don't
have to copy them line-by-line?
Is there a way I can nix-build it
without having to modify anything global (eg pkgs/top-level/all-packages.nix)?
Where is pkgs/top-level/all-packages.nix?
One option is to clone the nixpkgs repository and then build the hello package recipe provided in that repository:
git clone https://github.com/NixOS/nixpkgs
cd nixpkgs
nix-build -A hello
Doing it this way, you don't have to modify all-packages.nix, because it already has an entry for hello. If you do want to modify all-packages.nix, you can find it in the nixpkgs repository that you cloned. Just take the path of the repository you cloned, (e.g. ~/nixpkgs) and add pkgs/top-level/all-packages.nix to get the path to all-packages.nix. You can see a copy of that file here:
https://github.com/NixOS/nixpkgs/blob/master/pkgs/top-level/all-packages.nix
When you start building your own software that is not part of nixpkgs, you might chose to write your own default.nix file in your own repository and put a line like this in there to import nixpkgs, using the NIX_PATH environment variable:
let
nixpkgs = import <nixpkgs> { };
...
Note that I am running NixOs, not sure if my answer will valid for other (non-Linux) system.
Is there somewhere I can download the source files from so I don't have to copy them line-by-line?
You could browse nixpkgs, All source files are there.
Is there a way I can nix-build it without having to modify anything global (eg pkgs/top-level/all-packages.nix)?
David Grayson give excellent answer already.
I would love to add some information.
nix-build will looking for default.nix on the current directory and the build result will symlink named result on current working directory.
Another way to test if nix expression could build is nix-shell which also looking for default.nix or shell.nix on current directory. If the build success you will get shell prompt with your packages avaliable.
Both nix-build and nix-shell have -I argument that you can point to any nix repository including remote one.
For example, if I use nix-shell -I nixpkgs=https://github.com/NixOS/nixpkgs/archive/master.tar.gz -p hello, nix will download binary cache if exists or build hello using current master branch expression and give me a shell which hello is avaliable.
$ nix-shell -I nixpkgs=https://github.com/NixOS/nixpkgs/archive/master.tar.gz -p hello
downloading ‘https://github.com/NixOS/nixpkgs/archive/master.tar.gz’... [12850/0 KiB, 1404.5 KiB/s]
unpacking ‘https://github.com/NixOS/nixpkgs/archive/master.tar.gz’...
these paths will be fetched (0.04 MiB download, 0.19 MiB unpacked):
/nix/store/s3vlmp0k8b07h0r81bn7lh24q2mqcai8-hello-2.10
fetching path ‘/nix/store/s3vlmp0k8b07h0r81bn7lh24q2mqcai8-hello-2.10’...
*** Downloading ‘https://cache.nixos.org/nar/1ax9cr6qqqqrb4pdm1mpqn7whm6alwp56dvsh5hpgs5g8rrpnjxd.nar.xz’ (signed by ‘cache.nixos.org-1’) to ‘/nix/store/s3vlmp0k8b07h0r81bn7lh24q2mqcai8-hello-2.10’...
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 40364 100 40364 0 0 40364 0 0:00:01 0:00:01 --:--:-- 37099
[nix-shell:~]$ which hello
/nix/store/s3vlmp0k8b07h0r81bn7lh24q2mqcai8-hello-2.10/bin/hello
Where is pkgs/top-level/all-packages.nix?
There is NIX_PATH enviroment variable. The nixpkgs portion will point you to your current repository
$ echo $NIX_PATH
nixpkgs=/nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs:nixos-config=/etc/nixos/configuration.nix:/nix/var/nix/profiles/per-user/root/channels
My all-packages.nix is located at /nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs/pkgs/top-level/all-packages.nix

Extend $PATH variable in git bash under Windows

I'm trying to extend my $PATH variable in git bash (MinGW shell) by adding the following to the file ~/.bashrc
PATH=$PATH':/c/Program Files/maven/apache-maven-3.2.5/bin'
After I did this and restarted the bash it seems like that the $PATH variable was extended like expected:
$ echo $PATH
MANY_OTHER_PATHS:/c/Program Files/maven/apache-maven-3.2.5/bin
But I still cannot execute the programms in the given directory:
$ mvn
bash: mvn: command not found
What went wrong here? How do I extend the PATH variable correctly?
Here are two ideas.
You can have your path with double quote mark.
export PATH=$PATH:"/C/Program Files (x86)/apache-maven-3.3.3/bin"
Or, You can also make symbolic link for the directory.
ln -s "/C/Program Files (x86)/apache-maven-3.3.3/bin" ./mvnbin
export PATH=$PATH:/your-path/mvnbin
It works for me in mingw32 environment.
I needed to add something to my Git Bash path permanently each time I open it. It was Meld.exe path which can be added with:
export PATH=$PATH:"/C/Program Files (x86)/Meld/lib"
In order to execute this command each bash session, you need a ~/.bashrc file. Check if it already exists or create it using notepad ~/.bashrc or touch ~/.bashrc.
You can check where it is with:
echo ~
Open it and add the command that adds the PATH (first command in this response).
I hope you found this useful.
According to this SO post, you need to escape Program Files with quotes. git-bash $PATH cannot parse windows directory with space
Add PATH in Git Bash Permanently | Windows Only
Just in case you are still wondering how to add a path permanently in git bash here is the step-by-step process for Windows users:
Create .bashrc in user's root folder using the below command. It will open notepad and ask you to create the file, click yes.
notepad ~/.bashrc
Put the directory you want to add as below, for more than 1 items repeat the same format in next line:
export PATH=$PATH:"/c/folder/folder/"
Save the file and relaunch the bash.
Next launch will give you a warning like WARNING: Found ~/.bashrc but no ~/.bash_profile, ~/.bash_login or ~/.profile. but git bash will handle it by creating the required files.
SOME INSIGHTS
Git Bash doesn't fetch Window's environment PATH, it maintains its PATH separately in more like a Linux way.
You can run export PATH=$PATH:"/c/folder/folder/" in cmd to add a directory to path, but it will be only for the current session once you close the bash, it will be gone.
.bashrc is a shell script file that will be executed every time you launch a new git bash window. So you can add any type of bash command here. We simply added the export command to add our desired directory to PATH.

Setting path in Solaris Sparc

I have one python command file, i want to set it as a PATH in Solaris Sparc so that i can easily use my command from anywhere. For example the file name is abc.py and it contains abc --version to display version of file abc. So, after opening terminal i should only give command abc --version and it should display version of abc.
The architecture (SPARC) has nothing to do with the PATH which is more a shell thing but you do not tell what shell you are using.
Anyway, if you use a bourne style shell, i.e. not csh/tcsh, and you don't mind this to affect every user account on that host, you might add the wanted path to the PATH setting in the file /etc/profile.
When abc.py is located in your homedir, you can start it with ~/abc.py.
You need to call the file abc.py with abc.py, not abc (and have a shebang line which instructs the shell where it can find python).
When you want to start the file with ./abc, you can rename the file to abc (the shebang will tell it is python, not the .py), or introduce an alias:
alias abc="~/abc.py"
Using an alias can be an alternative for adding a shebang line:
alias abc="/usr/bin/python abc.py"
When you do not want to use an alias you can make a bin dir and put abc there.
I will add the shebang for you:
mkdir ~/bin
echo "#!/usr/bin/python" > ~/bin/abc
cat abc.py >> ~/bin/abc
chmod +x ~/bin/abc
mv abc.py bin/abc.py.old
Now change your login PATH with PATH=${PATH}:$HOME/bin in your .profile or .bashrc, and login again (or source the login script).

env: ruby\r: No such file or directory

I have Rails project. When I try to run any rake task or rails server it give me this error
env: ruby\r: No such file or directory
Could someone help me?
If you are working on a Unix / Mac, then this error is because you have incorrect line endings.
Here is a solution using dos2unix; you may need to install this program on your system. If apt is available, you can use sudo apt install dos2unix.
Set your line endings correctly, and have git manage how it handles them:
git config --global core.autocrlf input
In your directory you are going to convert all of the files by running:
find ./ -type f -exec dos2unix {} \;
This will cycle through all of your files, converting them. and solving the problem. Add your changes. Commit them, and you should be good to go.
You probably have edited ./bin/rake file and added \r at the end of first line:
#!/usr/bin/env ruby
begin
load File.expand_path("../spring", __FILE__)
rescue LoadError
end
require_relative '../config/boot'
require 'rake'
Rake.application.run
Make sure that after "ruby" you have only new line char.
Thanks to the comments above, I solved my server issue that was caused from cloning my group's github rails app and causing localhost:3000 to fail. I was just working on the backend from my fullstack app: ruby(-v 2.7.1)/rails(-v 6.0.3.4). And these 2 people's comments solved my error:
"For those of you who got "find: ‘dos2unix’: No such file or directory" error: sudo apt install dos2unix" – RealMan Jul 26 '17 at 14:59
"Note that that find command may be excessive... this point is arguable; it may well be fine, but it may be overkill in some situations. Another possible route (for step 2 in this answer) is git rm -r --cached . followed by git reset --hard HEAD... which is likely faster (if nothing else, it won't run dos2unix on files in the .git housekeeping directory!)... This has potential gotchas as well (probably quite fine if you're running from a "clean" checkout, though), but thought I'd at least mention it." – lindes Jul 13 '19 at 0:42
I kept getting this error and finally figured out how to fix it.
I made sure all the permissions on the files in my bin folder were
executable.
Run ls -lha in your current repository. You want each file to have an x at the end like this
-rwxr-xr-x.
To achieve this, you will want to run chmod +x <file_name_here> for each file in your bin folder, such as chmod +x rails, chmod +x bundle, etc.
Now when you run ls -lha you should see that they all have an x at the end.
Next, either in SublimeText, Atom or what ever text editor you have, you will want to check that you are not using Windows line endings. The \r character is something Windows uses. Unix just uses \n for a new line.
I use Atom so I went to the plugins section (Cmd + , on Mac) and then searched for line-ending-selector in the Packages section, and then went to the line-ending-selectors settings. Change your default to 'LF'.
You will find that at the bottom of files, Atom will tell you the type of line ending the file is using with a CRLF for Windows and LF for Unix/Mac. You want all your files to use 'LF'.
So in your terminal, open each file in your bin folder in Atom, by running atom ./bin/filename (such as atom ./bin/rake).
At the bottom you will see 'CRLF' or 'LF'. If you see 'CRLF', click on it and, at the top of Atom, you can choose 'LF'.
Cmd + s to save.
Do this for each. You are basically telling your file to strip all Windows line endings and use Unix line endings instead.
Once all files are edited, you should be able to run your rake or rails command.
Note: Sublime Text and Text Mate should have equivalents to Atom's line-ending-selector.
For macOS users
Step 1: HOMEBREW_NO_AUTO_UPDATE=1 brew install dos2unix
Step 2: git config --global core.autocrlf input
Step 3: find ./ -type f -exec dos2unix {} \; (in the repo you were trying to run your task on)
git add and git commit
You are good to go!
If none of the other answers works, try this:
git config --global core.autocrlf true
rails app:update:bin
I had the same problem on Windows Terminal, using WSL 2! I followed a post that recommended to install the dos2unix dependencie: sudo apt install dos2unix (Using apt package manager) and run other two commands:
git config --global core.autocrlf input (Set your line endings correctly, and have git manage how it handles them)
find ./ -type f -exec dos2unix {} \; (In your directory you are going to convert all of the files)
The git will identify a couple of changes, but you don't need to commit it. I just made a git restore . , remove node dependencies rm -rf node_modules and download it again yarn install.

Portable erlang

Is there a recommended way to "bootstrap" an Erlang distribution? I would like to run erlang on the bunch of machines where I do not have root elevation nor development tool-set (no compilers etc ...) . My thinking was to pre-package (on the machine with the same architecture) as much as I can before. What are the minimal requirements for an usable Erlang environment?
You can use the different erlware tools. Using the latest Sinan, you can even create a standalone release with the erts bundled.
Ok, I should have read it before ... (from INSTALL.md)
* Install using the `release` target. Instead of doing `make install` you
can create the installation in whatever directory you like using the
`release` target and run the `Install` script yourself. `RELEASE_ROOT`
is used for specifying the directory where the installation should be
created. This is what by default ends up under `/usr/local/lib/erlang`
if you do the install using `make install`. All installation paths
provided in the `configure` phase are ignored, as well as `DESTDIR`,
and `INSTALL_PREFIX`. If you want links from a specific `bin` directory
to the installation you have to set those up yourself. An example where
Erlang/OTP should be located at `/home/me/OTP`:
$ ./configure
$ make
$ make RELEASE_ROOT=/home/me/OTP release
$ cd /home/me/OTP
$ ./Install -minimal /home/me/OTP
$ mkdir -p /home/me/bin
$ cd /home/me/bin
$ ln -s /home/me/OTP/bin/erl erl
$ ln -s /home/me/OTP/bin/erlc erlc
$ ln -s /home/me/OTP/bin/escript escript
You can look at Wing3D for example.
CouchDB is another example. (Credit to tbikeev.)

Resources