0. Prerequisites

First thing you’ll need is to install thrift library that is used by Scribe to do all the networking communication. To build it you need to have boost C++ library installed:

1. Installing Thrift

Now we are ready to download and build thrift. I prefer to keep all manually built tools in /opt:

Please note, that make command needs root privileges (it installs Ruby bindings to the system folder).

You will also need to install fb303 library (it is used in all facebook/thrift related tools to do status/health monitoring calls):

2. Installing Scribe

Download Scribe from Sourceforge. Of course, you can use current development version from SVN.

3. Configuring Scribe

I’ve created the /opt/scribe/conf directory and copied examples/example1.conf configuration there:

Now we are ready to start scribe server:

4. Creating a test Ruby client application

First thing you’ll need is to gather all Ruby bindings into your app directory:

Do not forget to replace /path/to/downloaded/scribe with real path where you have extracted Scribe sources to.

And here is the test console Ruby script:

Woot, it works! Time to start creating your highly productive logging/data collection system.

P.S. On my machine I got the following output from this script:

To fix this issue, open the generated scribe/scribe_types.rb and replace require ‘fb303_types’ line with this:

First thing necessary to do is enabling UTF-8 support. If you create folders in Finder with extended characters (e.g. Russian, Japanese, etc.) iTerm will only show question marks instead of the real characters. To fix this you just have to set the LANG environment variable of the shell (put this into the ~/.bash_profile):

The next time you will start an iTerm session the variable will be set and unicode characters will be shown correctly.

A list of all supported UTF-8 locales could be retrieved with the command:

Next thing is to fix an input problem (from iTerm FAQ). If you can’t input Chinese/Japanese/Umlauts/Accents, put this into your ~/.bash_profile:

Ok, everything works now and all I want is to migrate my color scheme options from Terminal. I love dark color scheme with pastel ANSI colors theme:

There is a great feature of iTerm called bookmarks. You can create display, keyboard, and terminal profiles, and then add bookmarks with these profiles associated to them. Each bookmark opens a terminal window with specified command in it. By default iTerm creates a default profile with something like login -fp kpumuk (for my machine).

I’ve created a new display profile called “Pastel” with nice pastel colors which replace default ANSI colors in terminal (click to enlarge):

I think I’m not alone in my pastel wishes, so here is the script, which will import Pastel display profile to your iTerm, and will assign it to the Default bookmark. Please note: you should run this script from Terminal.app, and do not forget to quit from iTerm before running, because settings will not be applied in this case!

Hope, you will find this post useful. Fill free to post your comments and suggestions!

iTerm has a great support of AppleScript. Looking through the commands list I’ve found a really useful one: background image path. The last thing to figure out is how to create an image with the current host name, and I decided to use ImageMagick. So, let’s do it!

First, you need to install ImageMagick. You can use binary distribution, but I’d prefer the MacPorts version:

Let’s create a test image:

Here is the result and it looks pretty good!

Ok, now the most interesting part. I’ve created a bash script in ~/bin/ssh, which sets the background before calling ssh and clears it on exit. I suppose you have this folder specified before /usr/bin in your .bash_profile:

Here is the script:

Let’s try it:

So now I can stop shaking with fear when doing some destructive things in console — I know which host exactly I’m trying to kill. I hope you will find this post useful.

Updated 05/01/2009: Updated on_exit function to fix bug with scrolling after ssh session being closed.

Updated 05/03/2009: Added some variables to customize how host name will look.

First thing I want to mention is gems documentation. Every gem in system has powerful documentation based on rdoc, built during an installation procedure. The easiest way to view these docs is to run gem server and open http://localhost:8808/ in browser:

You will be able to review which gems has been installed, and check their documentation.

I’m pretty sure you have lots of gems with the same name, but different versions. You could use the following command to remove all gem versions except the latest one:

That’s all. You have only latest gems installed. But wait, what about Max OS X users, who have pre-installed Ruby? They (actually we) are in trouble, because…

Fixing problem with gem cleanup on Mac OS X 10.5.x (Leopard)

On Mac OS X when you will try to use this command, you will hit following error:

That’s common problem and I will show how to fix it. First, run following command:

You may now uninstall the offending gem with:

So, in general

will get you the location of the gem, and

will uninstall the gem (found here.)

I think it was pretty nice and successful, because I got many questions (really, really good ones), and instead of one-hour introduction we spent more than two hours in discussions and talks. I was enjoyed.

Below you could find slides from my talk (in English) and video recorded during session (in Russian).

Scribd Architecture Overview

The video has been splitted in two parts.

X1 Tech Talks #1: Scribd Architecture Overview, Part 1 from Dmytro Shteflyuk on Vimeo.

X1 Tech Talks #1: Scribd Architecture Overview, Part 2 from Dmytro Shteflyuk on Vimeo.

In this short recording you could see all guests who took part in this event. Many thanks to all of you, guys.

X1 Tech Talks #1: All Guests from Dmytro Shteflyuk on Vimeo.

Hope, you will enjoy this talk too. Your comments, suggestions, and ideas about upcoming talks are welcome.

Feedbacks on this talk:

• Развлечения айтишников: tech talks
• X1 Tech Talks #1: Архитектура Scribd

Take a look at the following example.

Here we have an array of fruits and we need to calculate a frequency of each fruit. As you can see, in the line 3 we are initializing frequency value to 0 if there are was no fruit with this name before. We can simplify this code:

In line 2 we are creating a new hash, which default value is 0. This means that if we would try to retrieve value for a non-existing key, 0 would be returned.

Let’s check another example:

Here we are collecting indexes of each fruit in the source array. But now we can’t just create a new hash and pass [] as the default value, because all keys in this hash will refer to the same array, so in result we will get an array [1, 2, 3] for each fruit. So let’s try the following:

In this case we are creating a new array object for any non-existing key, that was accessed. So

will return [] and save it in the hash. When you will hit this key next time, previously created array will be returned.

You can pass a block to Array constructor too. For example, you need an array with 10 random numbers:

You can simplify it using map method:

But you can do it even easier:

Next Memo will cover managing Ruby Gems, so stay tuned.

It’s implied that you know what is Git itself and how to work with it. Below you can find some advanced tricks, that should be helpful for you (at least they were helpful for me).

As you may know, to remove local branch git branch -d branch_name command is used. But if this branch was pushed to the origin repo, it will never be deleted there. To remove a remote branch use the following command:

where branch_name — the name of your branch.

When you are working in a team, everybody can create and push their own branches, which will be retrieved from the Git server during git pull or git fetch. If a branch being removed from the server, it will remain in your local repo forever. To clean these stale branches up, use this:

Also don’t forget to take a look at the git_remote_branch command-line tool, which makes work with remote branches as easy as it is possible.

Have any questions? Ask me and I will answer you in following Memo- posts.

And then I’m googling on how to install these gems. It’s time simplify my life and post commands here.

Installing the ruby mysql gem on Mac OS X and MacPorts

Installing mysql5 from MacPorts:

Now we can install mysql gem:

Installing the ruby memcached gem on Mac OS X and MacPorts

First you need to install memcached and libmemcached:

And then memcached gem:

If you have any questions that could be covered in this series — ask me in comments.

Here is example from fb:editor documentation:

First we need to include fb:editor tag to the page to force loading of CSS files:

Now we should define basic structure of the form:

BTW, we have several enhancements in the original layout: we have specified for attribute for the label tag (now if you would click to the label, corresponding input would be selected), and added ability to mark required fields with required text (it’s impossible with fb:editor).

Or even:

Almost 50% of this could be easily removed, and you will get beautiful and clear code, just right after you will understand what the hell is operator ??. Here is information from MSDN:

The ?? operator returns the left-hand operand if it is not null, or else it returns the right operand.

Easy and clean explanation. Let’s try to rewrite all these examples using this operator:

In case of value-types this approach will not work: you will get NullReferenceException. But do not worry — in this case we can use Nullable-types. Before:

After:

When you are using ?? operator, expression will be evaluated in the left to right order, so you can use something like this:

Of course, it looks more readably than the:

And a fortiori than:

Add it to your armoury!