Thursday, December 27, 2007

Automated Testing and Programming Skills

Joel Spolsky's lecture at Yale university makes some interesting points about the dangers of automated testing. Joel is good at latching on to the unintended consequences of latest trends, especially the kind that nerds with Asperger's typically overlook. He speculates that 'automatic testing religion' was what did Vista in. There is an interesting bit about the fate of testers without programming skills at Microsoft:
Microsoft has had a long term policy of eliminating all software testers who don’t know how to write code, replacing them with what they call SDETs, Software Development Engineers in Test, programmers who write automated testing scripts.
I can sympathise with this, having interacted with testers without programming skills myself. If Microsoft's policy were to be adopted in India, there would be quite a few unemployed testers. Joel's article is a helpful reminder of the dangers of overdoing automated testing. But I think the situation in India represents the other extreme - we have lots of testers without any real skills just trying to fill up bug trackers and warming chairs in head-count based projects. Automation is the last thing that concerns either testers or software companies in India. After the dot-com bust, testers have become an overvalued commodity in Indian software companies with clueless managements.

Wednesday, December 12, 2007

John Gilmore's test of a good programmer

John Gilmore suggested to me a test of a good programmer: one who has written a piece of software that at least 1000 people have downloaded from the Internet.

- Arun Mehta

Thursday, November 22, 2007

OLPC Laptop Application Virtualization

I was listening to a fascinating presentation (video)by Ivan Krstić, the director of security architecture of the One Laptop Per Child (OLPC) project. What struck me most was the extent to which virtualization is being used. Unless you have been living under a rock, you know that virtualization is a 'hot' area now. But this must be one of the few attempts to use it to provide desktop security (Incidentally Anti-virus tools routinely use virtual machine containers) The Bitfrost security model involves running each application in its own virtual machine container. As the Wikipedia entry explains,
Every program, when first installed, requests certain bundles of rights, for instance "accessing the camera", or "accessing the Internet". The system keeps track of these rights, and the program is later executed in an environment which makes only the requested resources available. This is implemented by a fully-fledged, container-based virtual machine.

By default, the system denies certain combinations of rights; for instance, a program would not be granted both the right to access the camera and to access the internet. Anybody can write and distribute programs that request allowable right combinations. Programs that require normally unapproved right combinations need a cryptographic signature by some authority. The laptop's user can use the built-in security panel to grant additional rights to any application.
Running each application inside a virtual machine is a scary idea, especially given the hardware configuration of the OLPC laptop. Krstic explained that the Linux kernel VServer patch had been extended to do OLPC specific tasks. Perhaps, the most fascinating piece of information that I learnt from the whole talk was the low overhead associated with such Linux VServer virtual machines. Quote from Krstic's presentation:
"The interesting thing about this, by the way is you know people are terrified of how are you going to do virtualization with 466MHZ CPU - turns out with Linux Vserver that the overhead you pay is 32 K per task struct, but there is 0% measurable CPU overhead with up to 65,000 Virtual Machines running ."
I have mixed feelings about this approach. Having done virtualization related work for the last two years, I find this approach fascinating;but, it could also start a disturbing trend of people opting for this approach instead of trying to solve security problems at a more fundamental level. But I got to admit that the clever Copy-on-Write approach of the VServer just blew me away. As Krstic said after explaining the low overhead associated with the virtual machines,
"I will let that sink in for about two seconds"

Friday, November 09, 2007

It's not as bad as the Sendmail configuration file

But the Apache web server configuration file is pretty close. I once heard one of the founders of the Apache project confess that, the only configuration file worser than it was the one used by Sendmail; I mean, that's not saying much, is it ? I wasted the whole afternoon yesterday wrestling with the VirtualHost directive. Finally, I threw in the towel and got it working in five minutes with twisted web. It's httpd.conf that's twisted. All I wanted was a couple of web servers on my laptop to test my program. The worst part - it makes you feel so stupid for no fault of your own.

Saturday, September 15, 2007

The Glorious History of HR in Software Companies

Mitch Kapor's wife Freada was in charge of HR at Lotus in the early years. (As he is at pains to point out, they did not become romantically involved till afterward.) At one point they worried Lotus was losing its startup edge and turning into a big company. So as an experiment she sent their recruiters the resumes of the first 40 employees, with identifying details changed. These were the people who had made Lotus into the star it was. Not one got an interview.
- Paul Graham in a footnote to his essay News from the front

Tuesday, September 11, 2007

Extreme Pair Programming - Guy Steele and Richard Stallman

"We sat down one morning," recalls Steele. "I was at the keyboard, and he was at my elbow," says Steele. "He was perfectly willing to let me type, but he was also telling me what to type.

The programming session lasted 10 hours. Throughout that entire time, Steele says, neither he nor Stallman took a break or made any small talk. By the end of the session, they had managed to hack the pretty print source code to just under 100 lines. "My fingers were on the keyboard the whole time," Steele recalls, "but it felt like both of our ideas were flowing onto the screen. He told me what to type, and I typed it."

The length of the session revealed itself when Steele finally left the AI Lab. Standing outside the building at 545 Tech Square, he was surprised to find himself surrounded by nighttime darkness. As a programmer, Steele was used to marathon coding sessions. Still, something about this session was different. Working with Stallman had forced Steele to block out all external stimuli and focus his entire mental energies on the task at hand. Looking back, Steele says he found the Stallman mind-meld both exhilarating and scary at the same time. "My first thought afterward was: it was a great experience, very intense, and that I never wanted to do it again in my life."
- Guy Steele on a hacking session with RMS in the 1970s.

Saturday, August 18, 2007

Beautiful Code - The Paris Hilton Connection

The odds of finding truly beautiful code in most production systems seem to be on par with the odds of finding a well-read copy of IEEE Transactions on Software Engineering in Paris Hilton’s apartment. Furthermore, enterprise software, which represents the bulk of code in existence, deals mostly with forms, reports, etc., which – one might argue – seldom require much thinking or cleverness.
- Albert Savoia in a post on O'Reilly's Beautiful Code blog

It's not just enterprise software - you find bad code in lots of startups, where there is an exclusive focus is on getting things done rather than doing it correctly. In fact, even lots of open source code bases are full spaghetti code. A great hacker even recommends FreeBSD code over Linux code.
Sadly, a lot of Linux and Gnome open source is poorly written.
- Trevor Blackwell in response to 'How can I become a better programmer ?'

Saturday, June 09, 2007

Startups are not Democracies

If you are not one of the founders of a startup, you really need to aware of this. I wish someone had reminded me about this :-( .

Startups are not a democracy. Want a democracy? Go run for class president, Bueller.
- From Mark Fletcher's 15 Startup commandments

The Dirty Secret of Blogging

That's one of the dirty little secrets of blogging: the discussion in the comments are often better than the original entry. - Jeff Atwood

Simplicity, a feature

Recently, at work, I had to use the XML-RPC.Net library and couldn't help smiling at this tongue in cheek answer in the FAQ section.

Why use XML-RPC instead of SOAP?

If your clients and servers are all running in the .NET environment there is no point in using XML-RPC: .NET provides excellent support for SOAP and XML-RPC doesn't have any features not provided by SOAP (other than simplicity).

Tuesday, March 27, 2007

Why Developing Software for Indian Companies is Not Easy

I was reading an article Indian IT biggies losing out to MNCs, when I came across a very interesting and insightful comment by Chanakya. Yet another instance where a comment on an article is far more interesting the article itself. A couple of years back, I quit my job since I couldn't not handle the irrational pressures involved in developing for certain Indian customers - I quit even without having another job in hand. I know of a very idealistically motivated software company that had created a great ERP product for the SMB market - the last I heard, they have stopped trying to sell to Indian customers and were focusing their sales efforts mainly on the US and Europe. It's not for want of idealism, or lack of technical competence, or domain specific competence that you find very little serious software development work targeted at the Indian market. Most of the the fine folks who have tried are suffering from Indian IT market burn-out, for the reasons for which are so eloquently explained by, Chanakya:

It is very true that the Indian IT companies are failing to corner big deals in the domestic market. Both the IT companies and the domestic clients are to be blamed for that.

Let us first start with the Indian clients. Their IT awareness is quite poor. Their business processes are also not very much IT-dependent. In fact most of them view computerization as an additional burden - a sort of necessary nuisance. The business processes of those companies are so non-standard that it becomes rather difficult for them to get the real benefits of IT. Today's well evolved softwares require clearly defined and standard business processes and, most important, discipline as pre-requisites. For example, to get the optimum performance out of a Merc you need proper roads and disciplined traffic. Otherwise its high costs of acquisition and maintenance seem to be unnecessary burdens on the owner.

With the exception of banks, telecoms and a few Indian pharmaceutical companies, almost all other Indian companies across the sectors have those inherent problems. So when they do not get tangible returns from their IT infrastructure they tend to pay substantially less for the same. That is the root cause of all the problems. The Indian IT companies focus mainly on the application development side with a distinct bias for manpower intensive projects. For them the earning per employee - or 'billing rate' in the industry jargon - takes precedence over everything else. Indian clients do not pay more than one-third of what an international client would pay. Add to that the inevitable delays associated with the domestic projects due to various reasons like unavailability of proper infrastructure, stiff resistance from the middle and lower management levels who see IT as a source of extra work load, erratic payment schedules etc.

Instead of staying away from such messy, low-profit (often loss making), difficult to manage projects the Indian IT companies - with the exception of Infy - have been systematically exploiting the Indian clients for training their freshers. The domestic projects are known as visa factories. Because of the strict regulations against import of cheap labour in most of the developed countries, visa / work permit applicants need to have a couple of years of experience as a pre-requisite. How do the fresh engineering graduates get that experience? Mostly through the domestic projects!

So how the MNCs are making money out of the Indian clients? Hardware supply is the key. They make enough money from hardware components - like storage devices, PCs, servers, routers etc. - to be able to cross-subsidise the application software. Indian clients - thanks to their great IT awareness - are very reluctant to pay for something intangible which they cannot touch and weigh! Software is such a component. Hardware, fortunately, does not suffer from that bias.

Before I end this post I want to share a funny but real story of an Indian client with the readers. Around ten years back one of my friends used to work as a free-lance developer. One of his many clients was a typical Lala-type trader. After making big money in trading, he wanted to add some class to his business. So he wanted my friend to develop a software for his company. After toiling for a couple of months and bearing with the paan-chewing, gaali-spewing Lala he finally delivered the software in the form of two floppy disks. The installation and the subsequent setup went off smoothly. The bill of twenty thousand gave the trader a big shock. "Bees rupiya ka floppy de ke aap bees hazar rupiya maang rahe hai? Sharm nahin aati?", he said. It proved to be rather difficult to convince him. My friend tried to educate him first. Then only could he hope to convince him! Finally he gave up. He settled for five thousand. Last couple of years the same friend of mine is in US. He is still a free-lancer. He never goes near any client whose skin colour is even vaguely brown.
BTW, don't take that last line too seriously, I am pretty sure he just forgot to add a smiley ;-)

Open Source Development - The Perfect is the Enemy of the Good

Subversion developers Ben Collins Sussman and Brian Fitzpatrick in a talk titled How Open Source Projects Can Survive Poisonous People (And You Can Too) (Google Video) stress the need to avoid analysis paralysis . One of the ways paralysis is caused, explain the developers, is the attitude of perfectionists that makes them indulge in endless technical discussions on mailing lists - this includes seemingly interminable discussions on even minor details; The 'Painting the Bikeshed' analogy by Poul-Henning Kemp illustrates this point. As Ben and Fitz put it in the presentation, "The Perfect is the Enemy of the Good".

I wonder if that's the reason they have put up with WebDAV for so long ;-)

Sunday, February 18, 2007

Dreaming in Code - Andy Hertzfeld Quotes

There are some great quotes from Mac hacker, Andy Hertzfeld in Scott Rosenberg's book Dreaming in Code.
  • "My style is to get something going really quick and then turn it into the great thing that is the reason you're doing it. You're not working to have it be run-of-the-mill. You're working on it to do something great. But you need to get it started! The key is getting exciting work going; the rest of it will take care of itself. You're sparking off each other - a virtuous cycle - once you're doing the thing you're there to do."
  • "There's no such thing as a typical software project. Every project is different."
  • "And in this meeting, we kill the snake - we don't just make plans to kill the snake."
  • "I'm the kind of developer who likes to throw lightning rods around. To make a great program there's got to be at least one person at the center who is breathing life into it. In a ferocious way."
  • "My way of writing code is, you sculpt it, you get something as good as you can, and everything is subject to change, always, as you learn. But you climb this ladder of learning about your problem. Every problem's unique, so you have to learn about each problem, and you do something and get a better vantage point. And from that vantage point you can decide to throw it out."
  • "Code is cheap. But often it tells you what to do next."
  • "We all need more glory as designers - to show we can design another great thing. Everybody who has a first success, especially when it's young, wonders: Was it luck, or was it skill ? Well, it's a little of both. If you can do another really great one, it shows the world something."

Monday, February 12, 2007

Mary Lou Jepsen - Inspiring

At age 29, Jepsen found herself suffering from blistering headaches, confined to a wheelchair, and sleeping 20 hours a day. She was just about to drop out of school when an MRI revealed a tumor on her pituitary, a small gland at the base of the brain central to hormone production. She underwent surgery to have the tumor removed and emerged from the ordeal ready to move on with her life. “There’s a stigma when you undergo brain surgery: are you still smart or not? So afterwards I tried to challenge myself to find out.” She finished her Ph.D. in the next six months and then cofounded MicroDisplay, a Fremont, Calif.–based company that manufactures liquid-­crystal-on-silicon chips for high-definition TV displays. She left MicroDisplay in 2003, citing “creative differences” with its chief executive, but within days Intel was recruiting her.

Her health problems weren’t quite over, though. As a result of the operation, Jepsen’s body now makes no hormones, requiring a rigid schedule of twice-daily hormone supplements to keep her alive. Now that she’s a globe-trotting computer executive for the OLPC venture, the regimen can be tough to follow; last March she went into adrenal shock on board a plane, forcing it to make an emergency landing. (On the bright side, Jepsen reports that as a result of her hormone deficiency, she is unaffected by jet lag.)

That's Mary Lou Jepson, CTO of the 'One Laptop Per Child' (OLPC) project.

Writing vs. 'A Real Programming Job'

Financially, writing science fiction is to writing technical books as writing technical books is to having a "real" programming job.
- Leonard Richardson, co-author of Ruby Cookbook and Beginning Python,
in an interview

How Come ZeroConf is Not Yet Wildly Popular

I have always wondered why Zeroconf adoption has been so poor. Stuart Cheshire of Apple who pioneered this technology, briefly explains one of the reasons in a presentation introducing zeroconf(6:16-7:30) at the 2006 Emerging Telephony Conference.

I imagine a number of people are thinking, well, if it's so great, how come I haven't heard of it ? And, that's an unfortunate history of marketing mistakes - I called it Zeroconf, but Steve Jobs thought that was a dull name. So he called it Rendezvous which is a really good name. But, some other company thought so too and they had registered the trademark.

Then we had a couple of Years of the marketing people and the lawyers doing their stuff. And meanwhile we had a technology without a name. So, in that two years every single printer vendor adopted it. But they didn't have a name to call it, which is why you won't see it on the box. But if you have a printer on your network, an ethernet printer, or a 802.11 wirless printer that you bought in the last couple of years, then it has bonjour in it. And, on the Mac you go File->Print and it's right there in the print dialog; No setup assistant, no wizard. On Windows we can't make it quite that easy. But you can download from and just run the setup assistant and it will find all those printers on the network, and configure them for windows for you.