In Pune, India for a vacation.

Switched to boradband from dial-up and was trying to work around the router BSNL provides with the connection. It says it supports wifi, and it does so decently.

BSNL engineer usually will create a dialer on your machine and charge you Rs. 600 to enable auto connect and "multi user" support. Multi user is nothing but allowing multiple machines (wired or wifi) to connect simultaneously to the router. Since it is dialer-based only 1 machine can dial and access Internet.

Enabling all this is like eating Goober Pie in Kramerbooks & Afterwords Cafe in DC (BTW, its amazing and a must try!). In short, very few changes are needed from the factory settings. Follow this to enable auto-connect and wifi on any of your BSNL dsl modem cum routers. Mine is DNA-A211-I.

1. Go to the default getway, most likely 192.168.1.1. If you are unable to get to it, feel free to reset the router to the factory setting by a small button at the back of the router.

2. Login using admin/admin, unless you have changed it. (You should!)

3. Go to "Advanced Setup" -> "WAN". Click on the "edit" button on the entry with values 0/35 for VPI/VCI and PPPoE for Protocol.

4. Select service category "UBR without PCR". Click next.

5. Select "PPP over Ethernet (PPPoE)" and Encapsulation Mode as "LLC/SNAP-BRIDGING". Click next.

6. Type username and password for your BSNL connection, "dataone" as service name and "AUTO" as Authentication Method. Check "Dial on Demand" and set 30 minutes (or whatever) as timeout. Check "Retry PPP password on authentication error" and "Bridge PPPoE Frames Between WAN and Local Ports (Default Enabled)". Click next.

7. Check "Enable WAN service". Click next, save and save/reboot.

The router should come up and have wifi enabled without having to provide BSNL password, ever.

All I wanted was a minimal root fs with development tools (gcc, make, ...) for UML. This is especially important when, say, you want to test new file system tools which snapshot a directory, and this is only possible when the tools are available inside the root fs. There are many root fs elsewhere on the web, but none with development tools (and rightly so, since for most cases you won't need those).

So created a VM using VMware Workstation and bzip2'ed the '/' without /proc, /tmp, and /sys and manually created those directories outside of the VM. Also chopped down the unnecessary things on startup using chkconfig(8) and modified /etc/inittab to give 2 xterms on boot. Root password is 'guest'.

Have uploaded it on my homepage: http://people.cis.ksu.edu/~gud/tools/suse11.1-rootfs.bz2

Saw this excellent Mira Nair debut movie Salaam Bombay!, and was wondering if the lead actor, 11 year old character called Chaipau (meaning tea and bread, in the movie hes working as a runner distributing tea for a tea stall) playing a street kid in Mumbai is really an actor. My suspicion was right, hes actually a street kid as like many other kids in this movie and no scene was shot in a studio.

I like such movies, non-fiction, real-life, albeit harsh more factual. Although I don't admire the intent of Mira Nair, as with this and noticeably in Monsoon Wedding, of showing only a bad picture of India. I am not contradicting myself, but there are things shown in her movies which are not needed and don't contribute to the story line. May be they are to be taken objectively or esthetically, or I'm missing the point. Nonetheless, Salaam Bombay! deserves all the accolades it has received.

(Chiller room is the term in the movie for a Government run center for development of children, which is shown as no better than a prison.)

A friend of mine asked what am I gonna do on my 1st anniversary here at VMware. I jumped and said, "I'm gonna clean my whiteboard da!"

Theres a reason to it. Reason being it still contains tidbits of things dating back to June last year! Small things...things explained to me, things discussed with team-mates, things discussed with friends, and some things I noted down myself when working on projects..scribbled in blue, black, red and green. Some failed, some worked! As I wipe things off, I can recollect the time of the day it was written, the context and with whom. As if all done just yesterday!

Cleaned it be for the new things to fruition.

One of my friends asks in his blog:

"If you could foresee the next two minutes in your life, would you do things differently than what you otherwise do?"

Foreseeing something is, IMO, helpful iff the outcome is alterable. But is it even possible?

Saying "..foresee the next two minutes.." is ambiguous to say the least. It could mean 2 different things: 1) what I'm foreseeing is one possible outcome, OR, 2) it is _the_ outcome. In case of (2), unless you're in a gambling business or in a life-threatening situation, foreseeing is mostly useless (and even harmful!). We'II come back to this later.

As for the case (1), seeing a possible outcome assumes a sequence of, say 'n', events e_0(0), e_0(1), ..., e_0(n-1), within those 2 minutes, where e_i(x) is causally related to e_j(x-1); i < j <= x <= infinity, and 'n' belongs to [0, infinity]. e_0(n) happens after e_0(n-1) and is foreseen. Now this is just one possible sequence of events and there could be infinite sequences and for any value of 'n'. E.g., there could be another event sequence e_0(0), e_3(1), e_2(2), ..., e_999(n-1), e_7(n); in this case outcome is e_7(n). To simplify the calculations lets assume 'n' tends to infinity (in other words, substitute 'n' for infinity) and at most there could be 'n' events in any sequence. This also means that there are at most 'n' possible outcomes, e_j(n). Also this means that there are 'n!' events in all. All the possible outcomes are equally probable. Now the question is which outcome would you foresee? For the outcome you see depends on penultimate event and a series of events before that, which are causally related and directly related to the event that is going to happen next.

For example, consider a set of colored water guns. I'm to pick any one and shoot on a wall in front of me. The outcome is the color on the wall. If we apply the aforesaid theory to this, I can foresee a color on the wall. Now, in case (2), no matter what gun I pick, I'II end up spraying the color I forsaw. In case (1), chances are (since I did foresee) I choose the gun with the color that I forsaw. Now consider a case that the guns are correctly marked with the color they have. Now if I see a color, and I'm to change it, I'II cleverly pick the right gun and get the color I want on the wall. BUT, but the moot point is what I should've foresaw? The color that I cleverly didn't allow to appear on the wall, or the one that I did?

With this could we conclude that being able to change the future you saw kind of defeats the purpose of your clairvoyance?

Even if we assume we foresee an outcome that would happen if we don't try to alter it, our ability of altering the output greatly depends on the rate of change of events, time between e_i(x) and e_j(x+1); i < j <= x <= infinity. For example, such clairvoyance would help me if I'm a cricketer or a stock broker, but it'd hardly affect me if I'm a tea-leaf picker or a carpenter unless it is life-threatening. Same is true in case of (1), with an added danger that since we foresee an outcome, now no matter what we do, nothings going to change it. This has some obvious consequences, and more so if the clairvoyance vision is 2 years and not 2 minutes. We could think that all we can do is work towards the next outcome. But, wait..think. I'm a film maker and I foresee that my movie is going to be trashed, I can't completely stop working on this movie and take up next, it has to be done until trashed, but all I can do is...is nothing! I wouldn't find time to work on my next film, for the time has to be spent on things I already foresaw sometime back! And the output is unalterable! Things have to be done; and by me!

And with this could we conclude that foreseeing an unalterable future is but useless and prenotion of disappointment could do more harm than cheer of an oncoming success?

Heard about this startup - Animoto. Its simple, it takes your pics, allows you to select a music track and compiles a video out of it and claims no 2 videos are alike. Neat, huh? The short videos (30 sec) are free but the longer ones are not. The shorter ones give a trailer kind of teaser feeling, and are actually great. How far are we to see such features in (*gulp*) Gimp and Picasa? Beyond a lame musical slide show, ftw :)

Just wanted to give it a whirl, uploaded a bunch of pics - random mostly and some from our recent quick Ghirardeli ice-cream late evening sprint. What came out is this - http://animoto.com/play/ni0VJUBru0DwstY7dplAgA

So parents depart after a good (short) half a years stay, and things change. I was preparing myself to be bored again, but it turns out, weekdays are (enjoyably) busy with work and every weekend continues to be equally fun-filled :) I now have more friends in the bay area than anywhere else, and the best thing is most of them are my old friends from Pune. Hiking, games and movies and coding make up my weekends, which otherwise was just coding :P

Blogging becomes scarce, out of pure ignorance, and having exciting things to do make it not so exciting ;) Plus twitter is there filling the need :)

LSF, is a premier meet for Linux file system folks. Being invite only, its the only concerned people that attend or rather allowed to attend this workshop. This is its third year and mine first. Only 60 odd people on the attendee list. Odd. Every big name you can think of in the Linux kernel community was there, notable exceptions of Linus and Alan Cox, but they aren't file system major anyway.

Heres the workshop program: http://www.usenix.org/event/lsf08/. And there wasn't any scope for any and every discussion not to be enlightening. Being a primary note taker, I was all the more attentive than usual. Good for me. Will hang off my notes here, as soon as I get them compiled in a readable way.

LSF was co-located with USENIX FAST 2008 and a BoF was arranged on the last day of the workshop which was open to both LSF and FAST attendees. I found it very interesting, especially since it allowed interaction between people from academia and Linux kernel maintainers. We discussed many issues surrounding file systems and storage apart from taking a overview of the discussions at workshop.

I enjoyed my presence at LSF a lot, more than just the day full of technical stuff, its about meeting and interacting with the small esoteric group of kernel developers, whom I revere since my school days. And now its easy for me to convince my mom that its ok if I work more, if akpm works 60 hours a week at his age :)

Ok guys, pinch yourself if you want to, this is true. I'm thinking of joining a company in Redmond, called M$. I've already started preparing for the interviews. Megs tells me the most common question they ask, like any other company, is "why do you want to join M$?". I've come up with few possible answers. Here they are:

1. I'm a sinner, I think I deserve it.

2. I've heard lots bad things about your company, just want to check out if its really that bad.

3. I'm done with my life. I think suicide would be little too much.

4. Nothing else worked. Plans A-Y failed, you're my last resort.

5. I'm anti-social. Everyone hates me anyways.

Disclaimer: umm...aa..aaa..umm...a pinch of salt and a grain of humor.


Btw, feel free to suggest me more :)