To be able to develop and test on such a VM I generally require it to be available as if it is on the network (per example via SSH, or my browser). Assuming that you have installed VirtualBox and that you know how to setup a VM already, I’ll briefly explain the specifics of setting up a VM that you can access as if it were on the network:

• When creating the VM give it two network adapters.
  • Adapter 1 setup as “NAT”, this gives the VM access to whatever network your computer is connected. This includes the internet if you’re connected to it, which can be useful to download OS updates or applications inside your VM (I often have VMs running Ubuntu, and the NAT adapter allows me to use apt painlessly).
  • Adapter 2 setup as “VirtualBox Host-Only Ethernet Adapter”. This gives the VM an IP address within a private range, that only exists within your own computer.

When installing an OS on your VM, it is useful to give the it a static IP address for adapter 2. Having a static IP for each VM will allow you to connect to it using bookmarks instead of having to check what the IP of each VM is after it has started. Note that you do have to adhere to the IP range defined in the VirtualBox preferences for the host only adapter. You can find this information via the menu: go to “File” -> “Preferences”, go to the “Network” section and double-click on the host only adapter entry to get a popup where you can see/configure the network range and the DHCP server settings (you probably want to ensure that your static IPs cannot be given out by the host only adapter’s DHCP server).

I started by reading online resources on both editors, and buying and reading a PDF book on each editor from O’Reilly. The books I bought are: Learning GNU Emacs and Leaning the Vi and Vim Editors. I found both books informative and pleasant reads.

As I went on I found that both editors more or less have the same capabilities (and extremely extensible, Vim through Vimscript and Emacs through Lisp), and that choosing between them really is a matter of personal preference. I settled for Vim as my editor of choice, below are a couple of reasons why:

• I prefer Vim’s modal editing (It is possible to achieve this in Emacs as well, if you really want it…)
• Vim has less things that I don’t need than Emacs does (Emacs has a calendar and a set of mini games per example).
• One of my colleagues at work also uses Vim (and because we both use it, we can share Vimmy things, such as Vim scripts and macros)

Before you attempt such an upgrade yourself, be sure to make a backup of your Linode. If you have enough diskspace left with Linode, the easy way is to resize the image that you’re using for your server and then duplicate it. Alternatively you can download a copy of the image (which can be quite time consuming, depending on the size).

The Linode manager makes it easy to use different disk images with your Linode, so if you mess up then you can switch to the backup image and revert to the situation before the upgrade. Kudos to Linode for offering such an excellent service with such excellent tools!

The upgrade was relatively problem free, the only real issue that we ran into was that somewhere in between 2.1.2 and 2.3.4 the *_tag helpers (text_field_tag and such) have started behaving slightly different in relation to element ids. These helper methods now call the undocumented sanitize_to_id method, which cleans up the element ids by getting rid of characters such as [ and ] and putting in underscores instead. While this is a good change (the form_for helpers were already converting the field names to such element ids, so this change makes the helpers more consistent), it was unexpected (I have not seen any documentation referencing it, not in the 2.2 or 2.3 release notes). This changed a lot of ids in web forms that we were using, and it broke a fair bit of custom Javascript which we had to repair.

So, if you’re upgrading Rails from 2.1.2 or older to 2.3.4 ensure that you check your views for custom Javascript to ensure that any fields generated by the *_tag helper methods that are being referenced with Javascript are referenced properly.

Happy coding!

In my case I needed to access a PPTP VPN, the package that Ubuntu has to do so is pptp-linux (if you need another type of connection you should search for the word vpn in the package manager). Installing that package does not add options to the network manager. While you could configure the VPN by hand (by editing configuration files somewhere in /etc, depending on the installed VPN client), an easier way is to install a plugin for the network manager, and then use the network manager’s GUI to configure your VPN connections and to control use of to those VPN connections. For the pptp-linux the network manager plugin package is network-manager-pptp.

When you use the standard version of Ruby getting SQLite support is easy, you just install the sqlite3-ruby gem (gem install sqlite3-ruby) and you’re done (you need to have the SQLite libraries and dev headers installed on your system already of course). This does not work for JRuby because the sqlite3-ruby is a native gem (containing C code with extensions on standard Ruby).

So how do you get SQLite to work with JRuby and Rails? Get the gems for JRuby! Assuming that you have SQLite installed on your system, you simply need to install the activerecord-jdbcsqlite3-adapter gem (this will automatically try to install the activerecord-jdbc-adapter and jdbc-sqlite3 gems if you don’t already have them). In your Rails project edit config/database.yml and change the adapter type to jdbcsqlite3 and you’re done

There are a number of activerecord-jdbc gems available (just type “gem search –remote activerecord-jdbc” in your shell to see them), including connectors for Apache Derby, MySQL and PostgreSQL.