Core Dump

Caving (or Spelunking) is the sport of exploring caves. Yosh and I recently had the opportunity to go on our first spelunking adventure. It was offered at the Timponogos caves in Utah for a group of five. It wasn't quite as dangerous and difficult as it sounds, but it was definitely more than just walking inside a cave. We donned our hard-hats with head lamps before entering the dark hole in the mountain. We were accompanied by a Ranger who knew what he was doing.

We had already taken the regular tour of the cave which in retrospect was dull in comparison to the spelunking experience. The path we took had a pretty low ceiling at many spots, sometimes so low we had to crawl in all fours. At one point we had to descend down the cave through a rock-climbing rope which was exciting. The cave walls were adorned with gorgeous calcite formations. At the end of our path there was a hidden lake (more like a pond really) which had a crystal clear surface. We all turned off our head lamps at this point and experienced the total darkness for about a couple of minutes. This is the kind of darkness that our eyes never get used to. No matter how long you stay in that darkness, it is impossible to detect anything. Experiencing the absence of even a single photon was quite unnerving. It was a lot of fun and definitely worth trying. We are both planning to pursue more spelunking in the future, lets see how deep it takes us.

Picture Time:

All ready to go in

Posing with the ranger

The whole group

Time to do a Part 2 on my series of Favorite Ted Talks. I've expanded my repertoire since I published part 1 to include talks that are not directly science related.

Benjamin Zander on Music and Passion:

Biking into a lake is not quite the same as Biking around a lake or Biking near a lake, it is orders of magnitude more fun than both of them combined. If you are not familiar with the concept of Biking into a Lake, here's a quick intro:

1. Take your bike and point it towards the middle of the lake.
2. Start pedaling as fast as you can.
3. Go as far as possible into the lake.
4. Fall into the water (trust me this will happen)
5. Repeat.

The lake had a very mild slope which was very helpful for this art, but it was way too muddy to keep a steady biking pace. The bike tires would sink into the mud making it impossible to restart once we got stuck (and fell into the water). Trying to pull the bike out of the mud resulted in our feet sinking deeper into the mud.

After a few iterations of lake biking, Thad and I decided to go for a swim in the lake. We both managed to go deep enough in that freezing water to reach neck level water.

Can you see the tiny little heads sticking out of the water. We were hoping our bodies would soon go numb. But the water just kept getting colder as we went deeper, I could almost hear my skin cells screaming "Are you f-ing kidding me?". We decided to turn back once I realized I might panic if we kept at it.

One of these days, I'm gonna man up and go all the way.

The other day I was reading an article online titled "How an Intern Stole Nasa's Moon Rocks". I didn't give it much thought at the time and moved on with my life. But on the back of my mind I developed a sense of respect to the man who pulled that incredible heist. A couple of days later my wife sent me an email that knocked me off my ass. The basic gist of the email is that her TA is Thad Roberts. For those of you who didn't read the article, he is the master mind behind the moon rock heist of 2002.

I could not contain the excitement once I found out that he is actually meetable (not a real word). Of course my wife managed to get me a little jealous when she wouldn't shut up about him. But I was never happier in my life with my wife's decision to take that Physics class.

Yesterday was the first time I got to meet the man in flesh and bones, in a "Physics" party for the students of that class. To be honest, I was expecting a charming, charismatic, adrenaline-seeking, egocentric, alpa-male and I was right up until the adrenaline-seeking part. He was everything the article described him to be, and a bit more. What the articles missed to mention was that he is actually a really nice guy who is down-to-earth and friendly. A regular guy with a regular life. I had a great time talking to him in the party and to my surprise it was probably the first time I actually enjoyed being at a party.

At the end of the party he recounted the experience of liberating the rocks. I call it liberating because he was only taking the used up rocks that were sitting in a metal safe (not even on display) that bear no scientific value anymore. The story from the horse's mouth was a little bit different from the one I read on the article. Apparently only 30% of the article is true and the journalist filled in the rest with some fluff to make it juicy. He did send me the original article that he personally approved for publication. But listening to the story from him was a lot more fun and interesting. I didn't ask him about a couple of questions about the story, but it was not the time and place, it'll have to wait.

He invited us to join an adventure club that he is putting together, I can't wait :).

Update: Thad has his own webpage with links to his new book (Did I forget to mention he wrote a book on quantum space theory when he was in jail?) that is due to be released soon and a blog with some pictures.

I recently got into a Ted.com watching spree and found some great talks (mostly scientific) by some of the top scientists in the world. The amazing thing about the talks is its simplicity and the engaging nature in which the speakers present the topics.

I have a lot of favorite talks and this is my attempt to collate the ones that impressed the heck out of me.

Brian Cox on CERN's Super Collider:

Whenever I write something controversial in my blog and contemplate about hitting the publish button, I think of the following XKCD comic.

Incremental string search is a feature that is available in almost all of the recent editors/browsers. It is also known as find-as-you-type in layman terms.

Typically a string search involves getting the search term from the user and matching it against each string in the input array to find the match. Incremental search tries to do partial matching as soon as a character is pressed. The results are narrowed down for each key that is pressed.

Conditions to satisfy:

1. When no character is pressed, the search result must contain all the strings from the input.
2. When a character is pressed, the search result must be the sub-set of the previously shown search result.
3. When backspace is pressed we include new strings from the input set to the search result which satisfy the new search term.
4. When a new character is pressed that doesn't match any of the strings in the input set, the search result must be empty.

Algorithm:

1. Initialize the array (or vector) of strings (in_array) which will be incrementally searched.
2. The search_term is a string which is empty to begin with. This string will grow/shrink depending upon users input.
3. 'cursor_pos' keeps track of the position of the current cursor.
4. 'pos' keeps track of the position of the character that will be checked.
5. 'possibles' is a stack that holds an array of integers. The integers in the array point to the matching strings in the 'in_array'. The bottom of the stack always has all the indexes of the "in_array". The top of the stack has the latest indexes that match the search_term.
6. loop until ESC is pressed.
1. Get a character and store it in 'ch'.
2. If the character is not a backspace or ESC add it to the search_term and increment cursor_pos.
3. If character is ESC break out of loop.
4. If character is back-space, remove the last character from the search_term, decrement cursor_pos (make sure cursor_pos is never negative). Pop the stack to remove the top most search result.
5. look_for the strings in 'in_array' that has the character 'ch' at location 'pos'. Add the indices of the strings that match, to a vector and push the vector into the 'possibles' stack. If no strings match the character, then push an empty vector into the stack.
6. Get the top of the stack and print the strings that are referenced by the indexes.
7. END_PROGRAM.

Algorithm Animation:

The animation below shows the state of the data-structures when the algorithm is running.

Code:

The algorithm described above was implemented in C++ using STL libraries and ncurses. 'ncurses' is a unix library for designing command-line interfaces. It is mainly used in this project to read the character as we type it, so replacing it with a windows equivalent will work just as well. The code is uploaded on GitHub for download.

Caveat:

• The goal of the code is to demonstrate the incremental search algorithm, it cannot be used as a drop-in library to include in a software.
• It doesn't capture every scenario of the incremental search. Eg: If a user deletes the first character in the search term the code can't handle it.

I had the most unpleasant conversation with an acquaintance the other day. He is a conservative person with thoughts that Gay Marriage is unnatural and probably a sin. I don't know why he felt it is necessary to share that with me, but I was dumb-founded by his arguments. I didn't really have the heart (or courage) to tell him that I'd probably vote yes to legalize gay marriage.

But later when I pondered about the subject of sexual choices, I realized I had no clue about them whatsoever. Of course, I have no problems with people being gay and two man or woman wanting to marry. I definitely don't think it is a sin, but what is the reason for a person's choice to be gay. Is it is a life-style choice? or Is it a genetic construct?

I've never had a friend or a family member who is openly gay, so I don't know their thought process. So I can't really agree to people's arguments that it is just a thought it can be overcome by prayer and discipline. What if it is more than that, what if it is not in the control of the person to choose his/her sexuality?

But this also begs the question, why are a lot of people averse to gays? Lets forget the religious people and consider the non-religious ones. Why are they so afraid of accepting the concept of homosexuality? Once again, could it be genetic construct that makes them unable to accept this concept? Are they afraid that a wide-spread acceptance of homosexuality might hinder the process of procreation thus resulting in the extinction of human race? Is there an evolutionary reasoning behind the hostility towards gays?

I did feel a bit shameful that I didn't speak up to voice my opinions that day.

I wrote a charming little utility for linux called getkey, that captures the keystrokes and displays them on the screen using OSD. Its called gekey and t has a couple of .c files and a hand crafted Makefile to go along with it.

But I wanted to use the auto-tools (autoconf and automake) to create a configure script and a Makefile so that my utility could by installed like any other GNU tool using the famous:

./configure && make && make install

AC_INIT([FULL-PACKAGE-NAME], [VERSION], [BUG-REPORT-ADDRESS])

AC_INIT([getkey], [0.1], [email-address])

# FIXME: Replace `main' with a function in `-lX11':
AC_CHECK_LIB([X11], [main])
# FIXME: Replace `main' with a function in `-lXtst':
AC_CHECK_LIB([Xtst], [main])
# FIXME: Replace `main' with a function in `-lxosd':
AC_CHECK_LIB([xosd], [main])

AC_CHECK_LIB([X11], [XOpenDisplay], [],[
          echo "X11 library is required for this program"
          exit -1])
AC_CHECK_LIB([Xtst], [XRecordCreateContext], [],[
          echo "Xtst devel library is missing. Xtst is required for this program"
          exit -1])
AC_CHECK_LIB([xosd], [xosd_create], [],[
          echo "xosd devel library is missing. xosd is required for this program"
          exit -1])

 AC_CHECK_LIB([libname],[func-name],[action-if-found],[action-if-not-found])

AM_INIT_AUTOMAKE([getkey], [0.1])

AUTOMAKE_OPTIONS = foreign

CFLAGS=-Wall -I/usr/local/include -DVERSION=\"$(VERSION)\" -DPROG="\"$(PACKAGE)\""
LDFLAGS=-L/usr/local/lib -lX11 -lXtst -lxosd

bin_PROGRAMS = getkey
getkey_SOURCES = getkey.c circ_list.c

bin_PROGRAMS = prog1 prog2 prog3
LDFLAGS=-L/usr/local/lib -lX11 -lXtst -lxosd

bin_PROGRAMS = getkey
getkey_SOURCES = getkey.c circ_list.c

My primary development environment at home is Linux (sometimes BSD). I do have Windows as a dual-boot (is it still dual-boot when you are running running more than two), but even when I am in windows, I develop in linux using andLinux, a variant of Linux that runs natively in a windows environment.

I use GHCi as my Haskell interpreter, since this seems to be the most standard compliant and the Haskell tutorial I am using is targeted around GHCi. My primary editor for writing programs is Vim and I use Github for storing my nasty little code snippets. I realize using git for this learning is probably an overkill, but I am a git n00b and there is no better way to learn git than just diving right in.

Even though I follow Learn You a Haskell(lets call this "LYH" in the future) tutorial, I do visit one other famous tutorial Real World Haskell everynow and then, when I am confused by a concept.

I think two posts without any code is probably an overkill for a blog series that is titled "Learning Haskell". So in my next post, I'll go over the first few chapters presented in LYH tutorial and share some code that I tried out.

Boy was I pleased to get this over with. Its been on my list of todo for a long time to give my blog a much needed overhaul. I hated the fixed width blog template with a passion. It would render just fine on smaller screens, but on the latest widescreen monitors, the white space on both sides of the blog was proving to be a criminal waste. Hence this new look and feel.

Learning haskell has been nothing short of an adventure for me so far. I love that is so different from the regular old C/C++. It's been a very long time since I've learned something refreshingly new. I usually spend about an hour or two on the nights to learn haskell and on those days I go to bed with a big grin and a satisfaction that you can only get when you've learned something truly refreshing.

I got roped into this by the myriad of comments found online raving about its simplicity and power. This one in particular by the creator of Haskell - (paraphrased) "Learning a functional programming language will make you a better imperative programmer" pushed me over the edge.

Now the reason to choose Haskell in particular is because it was notorious for being purely functional. It is one of the goals of the language to stay purely functional. Apparently it is really hard to write imperative style programs in haskell. Plus there is the much touted #haskell irc channel which has some of the friendliest haskell experts who are tolerant to newbies.

I started with the "Learn You a Haskell for Greater Good", it is a free online book for haskell that has some pretty pictures and presents the subject in a non-scary way (yes I scare easy). Haskell is so different from any of the languages I've known before, so it is only fitting that the tutorial itself doesn't fit the conventional programming tutorial's style. It doesn't waste time on how to write the infamous "Hello World!" program which IMHO is an incredibly useless excercise.

I've gone through the first few chapters so far and I decided to take it slow and try out every example given in the book (which btw, is a lot of code snippets). Since there aren't any exercise problems in this book, I decided to start solving Project Euler problems using Haskell. I am going to start recording my thoughts about Haskell as a series in this blog. I believe this series will keep me motivated to dedicate more time to learn Haskell and keep writing about my experiences.

void function_name(var1, var2)
int var1
float var2
{

}

void function_name(int var1, float var2)
{

}

void function_name(var1, var2)
int var1
float var2
{

}

I recently finished reading Richard Feynman's autobiography called "Surely You're Joking, Mr. Feynman!". I've never been fond of reading anything more than a page on my computer screen, but this was an exception that caught me by surprise. I found interesting insights in to the mind of a great Physicist. It had some fun anecdotes and some deep thoughts which inspired me.

Feynman is a nobel laurete who is lauded by Neils Bohr himself.

"Remember the name of that little fellow in the back over there? He's the only guy who's not afraid of me, and will say when I've got a crazy idea. So next time when we want to discuss ideas, we're not going to be able to do it with these guys who say everything is yes, yes, Dr. Bohr. Get that guy and we'll talk with him first."

"You have no responsibility to live up to what other people think you ought to accomplish. I have no responsibility to be like they expect me to be. It's their mistake, not my failing."

"Physics disgusts me a little bit now, but I used to enjoy doing physics. Why did I enjoy it? I used to play with it. I used to do whatever I felt like doing -- it didn't have to do with whether it was important for the development of nuclear physics, but whether it was interesting and amusing for me to play with."

"It was effortless. It was easy to play with these things. It was like uncorking a bottle: Everything flowed out effortlessly."

"We went into the bar, and before I sat down, I said, 'Listen, before I buy you a drink, I want to know one thing: Will you sleep with me tonight?'"

I recently started experimenting with Django as a learning experience. My friend and I created a nice little rss aggregator and decided to host it via DreamHost. Developing on my local computer was a piece of cake using the test server that comes with Django. But when we decided to move the setup to the hosting system, we faced our first obstacle, getting Django working on DreamHost. We followed some good tutorials here and here to get the initial setup ready.

But when we decided to setup a development version of our site (lets call it http://experiment.mydomain.com) we weren't sure how to configure Django so that we can have multiple databases and try some crazy things before pushing it to our production version. We did eventually figure it out, so here are some steps that might help fellow DreamHosters using Django.

First create a sub-domain for experimental purposes using DreamHost's administration panel. Make sure you enable FastCGI support for the sub-domain as well. I chose http://experiment.mydomain.com. Now ssh into your DreamHost server and there should be a folder for your new sub-domain in the home directory of your account.

I have my django installed under $HOME/django/src. The projects directory structure is as follows:

$HOME/django/main/projects/prj1 -- This contains my production code.
$HOME/django/devel/projects/prj1 -- This contains my experimental code, just copy over the files from the main directory into this one to begin with.

This way when I decide to launch multiple domains under my account, each one will have its own main tree and a corresponding devel tree.

Now lets configure the domains:

If you've followed Jeff Croft's Django setup tutorial, you'll have two scripts in your $HOME/mydomain.com folder. Which might look like the following:

#!/usr/bin/python2.4
import sys
sys.path += ['/home/amjith/django/src']
sys.path += ['/home/amjith/django/main/projects']
sys.path += ['/home/amjith/lib/python2.4']
from fcgi import WSGIServer
from django.core.handlers.wsgi import WSGIHandler
import os
os.environ['DJANGO_SETTINGS_MODULE'] = 'prj1.settings'
WSGIServer(WSGIHandler()).run()

#!/usr/bin/python2.4
import sys
sys.path += ['/home/amjith/django/src']
sys.path += ['/home/amjith/django/devel/projects']
sys.path += ['/home/amjith/lib/python2.4']
from fcgi import WSGIServer
from django.core.handlers.wsgi import WSGIHandler
import os
os.environ['DJANGO_SETTINGS_MODULE'] = 'prj1.settings'
WSGIServer(WSGIHandler()).run()

$ echo $PYTHONPATH
/home/amjith/django/main/projects:home/amjith/django/src
$ export PYTHONPATH=/home/amjith/django/devel/projects:home/amjith/django/src

$ django-admin.py syncdb

• Directory structure:
• $HOME/django/main/projects/prj1 -- Main tree
• $HOME/django/devel/projects/prj1 -- Devel tree
• Create a new sub-domain from DreamHost administration panel
• Create a new database from DreamHost administration panel
• Domain directories:
• $HOME/domain.com/dispatch.fcgi -- points to the main tree
• $HOME/experiment.domain.com/dispatch.fcgi -- points to the devel tree
• Modify settings.py in the devel tree to point to the new database
• Change the PYTHONPATH environment variable to point to the devel branch
• Do a syncdb in the devel branch:
• django-admin.py syncdb

My biggest complaint about Vim since I switched from Emacs is the inability to open a command shell inside the editor. I like to run 'make' or do some other mundane activity in the command line without having to exit my editor. It is possible to at least compile the source using make without leaving Vim. It also comes with the benefit of jumping to the errors in the code.

Commands to remember:

:make - This will run the regular old make in the command line for you.
:copen - Opens the quickfix window, to show the results of the compilation. Opens the window even if there are no errors in compilation.
:cw[indow] - Opens the quickfix window only if there is an error. If quickfix window is open and there are now errors, the window is closed.

Double click or press enter on the error in the quick fix window to take you to the corresponding error in the source code.

:cn - Goto the next error in source code.
:cp - Goto the previous error in source code.
:cfirst - Goto the first error in source code.
:clast - Goto the last error in source code.

For the sake of convenience I've mapped the keys 'cn' and 'cp' to :cn and :cp. So instead of typing <esc>:cn<enter> to jump to the next error I can just type cn and it'll take me to the next error in the editor.

It can be done through .vimrc by adding these two lines:
map cn <esc>:cn<cr>
map cp <esc>:cp<cr>

Regular expressions are the perhaps the most powerful tool to unix admins, programmers and most importantly Vim users. I am somewhat familiar with the Perl version of regex, but the regex used by Vim for its search and replace has some subtle differences which threw me off a bit.

Here are some examples that demonstrate the differences:

Input String: abc123def

1. Match one or more digits after 'abc'

Have you ever tried to multiply two 3 digit numbers using a pen and paper to find out that you aren't as glib at this art as you once used be? I had that experience a couple of years ago which shocked me a little bit.

Recently I had a very similar experience. I was writing a piece of code at my spare time and I was doing it in C. A little background about my education, I have a masters in Computer Science and I have been writing code for a living for the past 4 years. Back to my little project. Since C doesn't have the fancy STL libraries, I had to write the data-structure that I needed from scratch. I had to implement a Linked List and I was stumped. I knew the concept behind the linked lists and I know exactly the cost of every operation in the linked list, but when it came to writing one from stratch I was struggling to get it working at my first attempt. I realized how spoiled I've become by these fancy new libraries and languages. I did manage to get an ugly hacked version working in a couple of hours. It took me an additional couple of hours of coding and re-factoring to finally get it to a state where I was satisfied by my creating.

It was such a humbling and character building experience.

I posted my first full-length article in OSNews. It is a review of PC-BSD 7 which I am currently running on my desktop at home. PC-BSD is a fantastic OS and many thanks to Vijay for kindling my interest in this OS.

Writing the article wasn't too bad. It took me less than 3 days to do it. But after I finished it, I felt I could've written it differently. For some reason when I was proof-reading I felt it was too rigid, it didn't sound like my writing at all. I need to stop worrying about being correct and just blurt out what I think. (Note to self: Be relaxed when writing articles).

Now that it is up on the OSNews site, I can't help myself logging into the OSNews backend to check every hour to see how many readers I've managed to amass :). When I first published it this morning, it got a whopping 500 hits in 30mins. Looking at those numbers made me giggle with happiness. Its an amazing feeling when you know that someone is interested in hearing what you have to say.

1. A lot of people complained about the absence of AdBlock in Chrome. I don't care much for blocking ads, but it gets important if the websites are running Flash ads. The memory hog can be noticeably large if you have multiple tabs with flash ads. I'd at least like to have the ability to turn off certain flash elements on a page without having to kill the flash plugin for the whole browser.
2. The sub pages that get opened out of Google reader act like a child process. This makes it impossible to kill individual pages that the Reader has spawned off. So if a page in that group misbehaves, then I'll have to kill the whole group.
3. No provision to delete selected entries from history. I don't want to purge the history every time I want to delete just one entry.
4. Scrolling a page using the mouse seems jumpy in Chrome. I constantly find myself trying to locate my last sentence after a scroll move to continue reading an article. Smooth scrolling might help here.
5. The current method of page zooming (Ctrl ++ or Ctrl + Mouse Scroll) only increases the text-size. All other elements in the page (formatting, images) remain the same which makes the page look ugly. Firefox 3 does a good job of zooming the whole page (including the images). Here is an example:
Page View: Regular Size


Page View: Zoomed in Chrome



Page View: Zoomed in Firefox 3


6. I like the about:plugins and the about:memory. I feel like they should have a about:config for power users to tweak more.
7. Create exceptions to pop-up blocker. I don't want to click the pop-up block notification every time to see the pop-up window in a website that I trust.
8. A way to search my bookmarks will be a nice addition to the whole package.