For predictability, and to avoid messing up the user's system, I want the program's dependent gems isolated from any global gems the user/system has installed. So just distributing the program as a gem won't really work.
I know that Heroku's "toolbelt" CLI package used to be written in Ruby (though it's now Node.js, apparently); how did they do it? Is what they did a best practice?
I expect only developers to use the program (it's a release-management tool), so it's safe to assume/require the existence of a package-management tool like Homebrew on OSes that don't already ship with one.
The program is written to work on mostly any version of Ruby, so I can depend on some non-version-constrained Ruby package provided by the OS / package-manager, if that makes things easier.
Related
Recently I though about scaffolding a little CLI with Ruby, but was concerned about using it on a machine with no Ruby installed. I've searched for examples of popular CLI's and found that Docker CLI is built with Go language. I'm able to use this CLI on my computer with no Go installed. How can one build a tool that will not require you to install Ruby?
My guess is that there's a build process involved and it might be compiled to something present on most systems, like shell or smth. Sorry if this is a lame guess/question!
(note: this is not a detailed answer, just a summary of how it works)
Cli programs are just as other programs, there is nothing special about them.
Go is a compiled language - a program called "compiler" takes the go code and translate it directly to machine language, following the conventions imposed by an operative system. It becames pure 0 and 1, no references to anything else. The main advantage is that is self-contained, but you have to recompile it on every different architecture (32bit, arm processors, ...) and operating system (windows, linux, macos) - it's the operating system that take cares of redirecting input and output on cli.
Ruby, instead, is interpreted. There is a program called "Ruby interpreter" which translates your code to the appropriate machine code on the fly. It's a different approach, it's more "high level".
The advantage is that you don't need to recompile the code. However, the "Ruby interpreter" itself must be written in some compiled language.
I'm testing the bundling of applications including Ruby scripts using TideSDK, via the TideSDK Developer app.
I have several Rubies installed on my system, using RVM, and was wondering if there is any way of specifying which one should be used using a .rvmrc file or something similar that the TideSDK ruby module can understand.
Thanks.
how are you planning to run your ruby scripts? via Ti.Process? Otherwise, the TideSDK ruby module is what it is - there's no different versions you can use with rvm (Afaik).
Update/Expanded: I should point out TideSDK bundles its own Ruby - its not using the system ruby.
Otherwise, you can execute ruby scripts via Ti.Process (like any other script you would execute from the CLI). So, in your script, instead of #!/usr/bin/ruby you could do #!/usr/bin/env ruby. Of course, this wouldn't be very portable if you relying on rvm on target systems, but if you are doing a limited ditrib to known targets (ie: internally where you know everyone has a system setup with rvm) you could probably get something working.
I'm looking for a straightforward way to deploy a standalone Ruby-Tk desktop app on OS X. All existing options for deploying desktop Ruby GUI apps on the Mac (e.g. standaloneify.rb) assume that you are using RubyCocoa or MacRuby, and are tightly bound to Xcode/Interface Builder project structure; or, alternatively, that you are wrapping a simple script via Platypus that does not involve deploying the entire Ruby runtime. None of the other existing options that I've found--rubyscript2exe, ocra, crate--support the Mac and/or support Ruby 1.9.
I'm hoping that someone can point me to a shell script or Ruby script that does the following:
Tracks all dependencies/gems in an app.
Copies the Ruby interpreter and dependencies into an application bundle.
Runs install_name_tool on the Ruby interpreter and all dylibs to make them portable.
The application bundle can either be something created by Platypus, use a Cocoa stub launcher, or even a shell script--I know how to handle those. The problem I'm having is getting the Ruby runtime wrapped in a portable way.
There's a rakefile at http://weft-qda.rubyforge.org/svn/trunk/weft-qda/rake/rake_osx.rb that encompasses the tasks that I'm looking to achieve, but this rakefile is part of a larger build structure that I can't get running on my Mac. I'm not familiar enough with rakefile structure to refactor this into a simple script that I can run, either via "ruby rake_osx.rb" or even "rake".
If someone can suggest to me how to modify this rakefile to run independently, or point me to another example project, or even share some code, I would be grateful. The project I am working on is better suited for Ruby than Python (better library support in Ruby), but I am not going to invest time in it if deployment won't work. (Python has a much richer tradition of deployment tools than Ruby, something I'v never quite understood.)
This is a bit late after the fact, but in classic fashion, since I could not find a desktop app deployment tool for Ruby-Tk on the Mac, I wrote my own:
ruby2app
ruby2app is a simple command-line tool, inspired by similar Mac tools like py2app for Python, that bundles a Ruby script with the Ruby runtime into a standalone Mac application. It reads a basic configuration file to find the appropriate Ruby installation, compiles a small C program that embeds the Ruby interpreter, copies the entire Ruby installation into the application bundle, and re-links the relevant libraries with the main executable.
It attempts to provide for the Mac what ocra and exerb provide for Windows. Hopefully this will be of use to others.
I enjoy developing in Haskell, but am presently confronted with a challenge. At my place of education, I don't have administrative rights on my account. I want to debug/test while there. I need a way to run Haskell code without installing haskell-platform or ghci or anything else requiring of an administrative level of privilege. For instance, a web site which will run haskell code (abeit limited to not include file access of course) or a java utility (or standalone .app, though java is preferred due to the nature of the "parental controls"), or something else I haven't thought of. Source or compiled Java, Flash, and source code in Ruby or Python are preferred to native code. This is on Mac OS X 10.6, for reference.
You can install GHC without admin privileges, but you'll have to build from source. Check out the Building on MacOS X for more details.
Essentially, like any other open-source project, you can compile code and install it, generally, anywhere on your filesystem like into a folder in your home folder (I often use the ~/.local folder for that purpose).
As the linked page mentions, you can also use MacPorts and install it to any place you can write to. Once MacPorts is installed you can install GHC.
EDIT
As pointed out by Carl in the comments below, you don't need to build from source. Just grab the binary tarball from http://www.haskell.org/ghc/download_ghc_7_4_1#binaries.
In addition to all the other ideas, several companies will rent you virtual (cloud) linux servers for a few cents an hour. You have root on those and can install whatever you want, then freeze the image until you need it again.
Normally this might not be advantageous over a local solution if you can make one work, but a possible extra benefit would be that your work can stay on a single "computer" which you could access while sitting in front of any variety of modern PC that might be available to you on a given day.
An alternative not only for Haskell is http://ideone.com/
For the website option, TryHaskell will hardly cut it, it's way to limited.
codepad, OTOH, is more liberal, and should be of more use.
But honestly the bindist option is the best one, clearly!
If you truly can't install anything, then there's
http://tryhaskell.org/ -- like GHCi, but you can't load modules, which means not only that you can't use many standard functions (say, the functions in Data.List), but also that you can't use certain language features (like user-defined algebraic data types).
I also remember a hpaste-style site which executes its code -- and allows you to have private pastes -- but I can't remember it at the moment. Edit: I was thinking of http://codepad.org/ -- thanks #Mog
Try http://hiji.tinyrocket.se/ It is a haskell interpreter written using javascript.
http://ghc.io/ GHC.IO is a version of the Haskell interactive interpreter, ghci, that runs online in a web browser.
You just need to install homebrew, which you can do without root rights if you do so in your home directory. You can then brew install cabal-install
which will automatically install cabal and ghc, or brew install stack to install stack.
There is a web-based interpreter for Haskell at http://tryhaskell.org/. However, it may not provide enough functionality for your requirements.
I wrote a bunch of Ruby scripts that integrate different tools to create a "workflow for academics". A number of my friends have been very interested in trying this out, however they are held back by how hard it is to install or replicate my environment.
Although Ruby comes built-in with Mac OSX (the applications I integrate are all Mac applications, so unfortunately the solution can only work on Mac), I use a few binary gems, including Appscript (bridge to AppleScript). The problem is that this requires XCode, which my non-programmer friends do not have installed (and it seems a bit demanding to ask them to install 3GB of software to run a few kbs of code).
I tried freezing the gems by installing them with gem -i, into a subdirectory of the source (on github), and tried removing my entire central gems directory on my local computer - it worked fine, the scripts still ran, etc. However, when my friends tried running the scripts, they got error messages about missing library files etc.
So my two questions are:
* how specific are compiled binary gems? will they work on other Mac OSX versions? with other versions of Ruby? etc
* how can I find out exactly which binary files (located in central locations) a Ruby gem depends on, and package this with the source (and how do I tell the gem where to find these files in non-standard locations)
I know this might not be "standard procedure", and I should just package it as a gem and let my friends deal with the problems etc - but I'd really like to solve this, and I am sure there must be some way of doing this?
Thanks a lot
In my experience, compiled gems are not portable. They're dependent on the exact version of ruby and whatever other libraries they use. A better solution would be to have your friends install the "Command Line Tools" version of Xcode -- it's only 100-150 MB and includes everything you need to build gems (and other software). You'll need a (free) Apple Developer ID to download it though.
https://developer.apple.com/downloads?q=Command%20Line%20Tools