Chan 'dventure

OAuth and all the other authentication, authorization are full of jargon (and massive specs) and it’s really over complicated to the beginners who wish to learn about it. First thing we have to understand is what is happening in the big picture.

Early days scenario

When 3rd party applications required accessing data and features on Facebook, Twitter the approach was to give away the username and password of your Facebook/Twitter account to the 3rd party application. By giving away the username and password plainly to 3rd parties allowed unauthorized access of the account since everything can be accessed by the username password.

So what’s the real problem here?

Today’s cyber citizen has resources in the internet. These resources can be e-mail, user accounts, documents and many more. And almost always people needs to collaborate with other people and applications as well to achieve productivity in their tasks. But it was not necessary to give authorization to all the resources a user has to every application.

OAuth is a mechanism to use implement the above scenario. The user will have resources online. An access token will be generated per application for the necessary resources of the user. This token is valid for the particular user as well. By doing so we isolate the functions from authentication and allow unified resource usage.

I am looking into the V8 javascript engine. It’s the extremely fast beast that runs underneath Chrome making clock faster than any other browser. I am going to work on Jaggery.js these days and thought of implementing V8 to it rather than the traditional Mozilla Rhino engine but that story will go to another post. Today I am comparing both these engines to understand what one ups V8 over Rhino.

I am going to be mentioning a benchmark that was conducted comparing Rhino and V8 by a fellow blogger (I will be doing a comprehensive benchmark suite after I got V8 to execute via Java).

According to the benchmark suite V8 has a significant performance streak compared to Rhino. You can read the original blog post and view the git repo of the benchmark tests and see for yourself.

So what gives V8 this performance improvement?

JIT

Just-In-Time Compiler approach followed by V8 will compile javascript to machine code without intermediary Bytecode. But Rhino will have two layers of interpreters -Rhino and below that JVM. This might cause a around10x performance degradation.

Robust garbage collection

Implementing precise GC just like in Java, V8 gains advantages over here as well. Rhino also will have improvement in the GC arena since it runs on top of JVM.

Hidden class approach

In javascript there are no user defined classes so the executioner has to do type look ups as well. To reduce the overhead, V8 developers use a special concept of hidden classes. Where two similar objects instances will have a hidden class to improve performance.

function Human(height, weight){ this.height = height; this.weight = weight; } var h1 = new Human(40,150); var h2 = new Human(50, 170);

To understand better refer the below diagram -

Today was my first day at WSO2. I went to the 59 office with a bag of Hakuru since it was a Awurdhu party day at WSO2. Everyone was gathered and they had a friendly chat. It was getting bit awkward because I didn't know a single one of them and I started to panic since I couldn't find anyone of the mobile team. I thought they might have gone to the other office.

But soon after everything got sorted and it was a pleasant day. I got a brand new ThinkPad laptop from WSO2 work with. I installed Ubuntu 12.10 on it and ran into a decent problem after I got home (You must know this if you read my previous post). Other than that the laptop is great. It's a bad thing that Lenova doesn't make light laptops -compared to the one that I am using to write this post (My MacBook). 

At the first day of WSO2 I was given a ThinkPad T530 loaded with Ubuntu for my development work. It was a fantastic laptop and I got all my programs running and the download speed is great as well. I went home with the laptop and I wanted to check out the sound quality of the laptop as well.

I copied my Music folder from the NAS to the home directory and clicked the song hoping it would play. Unfortunately no sound was coming and I was worried whether there was a fault on the hardware. I searched Google and I found the culprit.

I didn't want to go through the painful advice offered at the end of the report -

Yuriy Vidineev, could you please test the latest upstream kernel available following https://wiki.ubuntu.com/KernelMainlineBuilds ? It will allow additional upstream developers to examine the issue. Please do not test the kernel in the mainline kernels archive directory daily folder, but the one all the way at the bottom. Once you've tested the upstream kernel, please comment on which kernel version specifically you tested. If this bug is fixed in the mainline kernel, please add the following tags:

kernel-fixed-upstream

kernel-fixed-upstream-VERSION-NUMBER

where VERSION-NUMBER is the version number of the kernel you tested. For example:

kernel-fixed-upstream-v3.7-rc1-quantal

But I was determined to find a better solution. Cause I knew that there must be a better solution. C'mon all of the other guys are using it :D. And it happens to be very simple -

/usr/bin/amixer -c 0 sset "Auto-Mute Mode" Disabled

I always wanted to monitor sleep cycles that I was utilizing :D. After having trying couple of methods (Spreadsheets, Self-made app) I found that I will never perform any of these tasks. 

But I really want to monitor sleep and increase quality of sleep. While pondering over the problem on my nice cozy rocking chair I hit the lightbulb. Why not use twitter?

It's something I use everyday and I occasionally send 'Good morning' tweets like below -

Good morning people... Got some work to be done today

April 7, 2013

If the tweet contain 'Good morning' or 'Good night' it will take in the time of the tweet and register it as sleep start/end time.

I am trying out coding and thinking about problems on my couch rather than on my desk. It’s kinda of creative and feels kinda of awesome. But the new postures (I feel bit like Buddha) are kinda of hurting my body. But let’s say it’s a nice little way to pass time while you are coding at your home - or preferably in your office.  

One of my colleagues was building a criminal database management system for an assignment for his coursework. There came a requirement where he wants to recognize a criminal of an image - something like they show in Television series.

It was a cool idea and he asked my help to build that component of the system. I looked into possible technical solution pre-made for our usage. And voila - OpenCV was right there to be used for the FaceDetection. OpenCV source itself has a special class that provided the features out of box. Though OpenCV is C++ based there exists a binding for Java via JavaCV.

Now the challenge was to plug in the OpenCV component to the PHP application. For this I build a small jar package that can be run via Command line. As for the inputs it will be using a file created by the PHP application. Then the results will also be written back to a file. The PHP application will read the output file and comprehend which criminal in the database is referred by the image.

Their were many challenges out there when building the small component - Due to lack of documentation and proper understanding of the subject it took many days to work out the solution. Also there are two different terms involved in the approach - Face Detection, Face Recognition. Face detection refers to finding if the image has human faces in it. Face recognition on the other hand features matching up against previously taken faces and figuring out which is the most matching face.

I’ll post the solution over at github as a gist in the coming week.

Sometimes we want to truncate all the tables in a schema or disable all constraints of a schema. It’s a pain to manually write all the queries. We can use the below Oracle features for our ease of use -

Disable all constraints

select 'alter table ' || table_name || ' disable constraint ' || constraint_name || ';' from user_constraints where constraint_type = 'R'

Truncate all tables

select 'truncate table ' || table_name || ';' from user_tables;

When it comes to database performance - it should be consistent and reliable. We had a setup where 60 users connected a MySQL database over network. During a certain process of the application MySQL database starts pulling in CPU power and reducing the database access speed due to the bottle-neck.

We had to do couple of things to solve this bottle-neck.

We first adjusted the couple of indexes on certain tables for our application SELECT queries. The query speeds were increased 50%.

Previously our MySQL database was surviving on Windows 2003 server edition. We switched it to RedHat to achieve the dynamic power of our virtualize setup.

We performed some tune-ups on the database - query_cache_type, query_cache_limit, query_cache_size, key_buffer_size, innodb_buffer_pool_size.

After the adjustment most of the performance issue was solved. After the above incident we understood that performance shouldn’t be taken for granted. And always we should understand the database we are dealing with.

I always wanted to tweet the song I am currently listening to. I have looked into various plugins and other ways to tweet out the #nowlistening. But couldn’t find one that matches my requirements and I thought heck... Why not roll my own solutions?

That was when I looked into AppleScript - it allows one to interact with Mac applications and obtain data out of them. For example we can extract the current track of iTunes and then get the title of the Song and the Artist.

In AppleScript how we talk to an application is defined by a tell block.

tell application “iTunes” # Define you logic end tell

Below is the basic interface of AppleScripting. You can Record and Run script from it.

As for the logic what we first need to do is to find out if iTunes is currently playing. If it’s currently playing then proceed into the program.

if player state is playing then # Logic over here end if

After that we can obtain the current track and then extract the name and artist. After that we call the System Events and we talk to the Notification Center. We then define the logic to press the button to press ‘Tweet’ button and then insert the string to the box and then sending the tweet.

tell application “System Events” tell process “Notification Center” click menu bar item 1 of menu bar 1 click button 1 of UI element 1 of row 2 of table 1 of scroll area 1 of window “window” keystroke “Now listening to “ & nameOfTrack & “ by “ & artistOfTrack & “ #nowListening” keystroke “D” using {command down, shift down} keystroke space end tell end tell

In the context of AppleScript there are API documentation for the applications available and they are call Dictionaries. You can access it by File -> Open Dictionary.

Then you can view all the available applications.

From there you can a documentation browser that will provide documentation for methods and properties of the application.

He learned about boredom. He learned why he was feeling bored and knew by inspiration what he needs to do to pull back the awesome spirits. How did Chan manage to be yet again awesome?

Bingo - He pondered into manga. He started reading a manga called ‘The One’ which was about a Modeling. The manga portrayed the life of a model and how the modeling industry works. And also a young girl’s enthusiasm to the world’s top model. Chan can’t remember his name though but he was portrayed to be very sexy and he also has a older brother who just looks like him. His older brother’s name was Eros. He was the character that appealed to Chan most. Cause he was a successful director in New York who was literally ‘AWESOME’.

He sometimes goes in disguise on the streets appearing like a beggar. Don’t know for what reason that is but he meet interesting people by wandering in disguise. So what does this has to do with Chan’s boredom anyway? After falling down into the pit of manga and getting into the story, mixing it with real life allows one to understand what he wants out of his own life. Manga allows one to learn more about one’s desires. His interests and what he likes to do.

Then he simply put the Manga away and thought aloof to the sky - who am I kidding -the Ceiling. Then Chan felt it - ‘I am Awesome. I am going to turn boredom to interesting turn point and make it awesome’. With the new change of attitude he got most of the things weighted in his mind sorted out. Fantastic right?

Now now I was a workaholic a while ago. But then I noticed I was making progress slowly. Slow progress because maybe I was tired or I didn’t have time to think out of the box. I was doing things according to standard procedure (for what ever it was :D). But then - I was always following my own philosophies right? So why not for work and fun ??? <Chan thinking>

Then I started going on adventures - venturing out on many things from travel, girls, drinks, food, racing, nightlife. You know what - then and there I discovered my self <Chan awakened>

The understanding is - You have 80 percent Fun and 20 percent Work. For a workaholic this might seem unfair since it’s not a balance ratio. Pareto principle supports my argument - 20 percent of things we do affect overall everything. May that 20 percent be Work and the other 80 percent be Fun. After all I live my life to have Fun everyday.

I first thought maybe I can make my Work fun. I was wrong because Work and Fun has a lining. The lining is - Responsibility. You have to be responsible for Work. But for Fun you really don’t have to be responsible for it. That’s why it’s Fun.

So let’s follow my current path and see how things will go.

Lock-in is the procedure followed by organizations to retain customers. The general paradigm is by increasing switching costs, you can increase the lock-in ratio as well. Another important point we need to understand is that when lock-in is involved in the industry new technologies will have to depend or interact with network externalities (Widespread technologies in the market) thus it increases barriers to entry (Porter's 5 Forces). 

To understand switching costs better let's take an example. If you are now using a windows laptop what additional costs would you have to switch to a Mac laptop (apart from the laptop itself).

Software

Certain hardware devices

User interface training

In a nutshell 'Switching costs' measure the extent of customer's lock in to a given supplier. Since we are talking about competitive advantage by designing products and services enabling switching costs would increase the competitive advantage of a firm. 

To further understand lock-in we can divide lock-in to 4 different sections.

Contractual lock-in

Technical lock-in

Economical lock-in

Emotional lock-in

Contractual lock-in

This type of lock-in is executed by means of a contract. First the firm tailors it's pricing and contract design to suite customers and add value to them in an extensive period. By doing so customers will sign a contract for a period of time where the supplier will have perfect monopoly. An example for this is Telecommunication industry offering contractual mobile smart phone plans for customers. The plans offered will enable customers to buy a smart phone for lesser price but it would be a contract for specific period of time. Customers are legally bonded towards the supplier lock.

Technical lock-in

This is a lock-in strategy that really plays well in the technology sphere. By using proprietary technologies the customer is locked into the supplier. The customer would have built a user interaction/ system dependent on the proprietary technology  so he will not have the power to switch technologies without major changes to the current procedure. An excellent example for this is Microsoft Office. If everybody is using Word and I will have to use Word to collaborate with others (Not so true after Google Drive's entrance). Since Microsoft had the first mover advantage it has gained massive market share and would dominate the market for long period of time. - Conclusion perfect lock-in. 

Economical lock-in

Costs are a major source of economical lock-in. If the supplier had setup fee for their system it would be a sunk cost to the customer if they are to switch suppliers. Also by enabling a long term pricing model will be convenient for the customer as well as the supplier. If vendor specific resources were accrued by the firm that would be also be considered as a sunk cost. Implementing RedHat linux servers can be viewed as a Economical lock-in. Since RedHat is open source technology the pricing model is subscription based and the cost of setup (Consultation) would be a perfect economic lock-in.

Emotional lock-in

Emotional lock-in is when the customer feels that they should always choose a particular supplier. The reason for this might be - Trust, Satisfaction, Preference. For a supplier to gain emotional lock-in they will have to engage with customers more and provide quality products and services. They will have to invest more into PR. In certain cases for example technology customers prefer the first-mover since that firm is the innovator of the industry. A fine example for this is Apple. 

The essential topic to understand when talking about Lock-in is path dependency. Choices of the future will be limited by the decisions we take today. The main reason for this is satisfysing (Choosing the first option that will work without looking into better options). 

Inputs and Outputs are the bread and butter of computer programming. You read from a hardware resource and you write to a hardware resource. Over the years there have been improvements on the Operative Systems on the hardware as to how IO should be managed.

Java has been a platform independent language that optimizes IO for the platform. The original IO api - java.io.* could not handle the change so the developers built the NIO API. New IO API provides -

Non blocking IO

Buffered IO (Bulk IO)

Selectors (manage multiple IO activities from single thread)

The NIO API was shipped with 1.4 JDK. Another update has been released to NIO and It's NIO 2 and it is shipped with Java 7. Below are the new features.

Asynchronous IO (Finally getting callbacks on IO operations)

Complete Socket-Channel facility (I am going through this)

Improvements to the file system (New java.io.File is java.nio.file.Path)

In this week I will be diving through the old NIO API and the newer NIO 2 APIs studying how things work and why the improvements matter to us. Teaser to some of the codes for your viewing pleasure -