Channel 9 - Discussions by Karthik

Coffeehouse - Google's future

Karthik — Sun, 03 Oct 2004 08:18:44 GMT

Manip wrote:

True, if they want to start from scratch.. but google MIGHT not need to. If they just want to develop Linux then they can get a head-start.

What makes you think that Google would want to release an OS based on Linux? That's not typical of Google -- so far, they've managed to innovate quite well by doing things differently.

Look at the people they've taken -- Joe Beda, for instance, used to work on Avalon, which is quite a strong core-infrastructure related area. If Google does anything at all, it would be something subtle and not something that's obvious.

Look at it this way -- Google's arms reach out to searches, newsgroups, emails, shopping and even into a significant number of desktops (in the form of Google toolbar/Picasa).

Who knows, they may come up with a new OS -- that you can remotely boot into off their network, without having to install it on your system. You never know. Even if they do, there is no reason it has to be Linux -- that would entail them to release the code into the pool, which they may not want to at this point. I would say even if they DO choose, it maybe something that resembles a FreeBSD-ish thingy.

Tech Off - What, the only technical stuff has to be programming related?

Karthik — Wed, 21 Apr 2004 15:52:49 GMT

What about us Computer Science types out here?

(the sometime-programming mostly algorithmic kind)

Coffeehouse - curly bracket placement thread

Karthik — Wed, 14 Apr 2004 20:23:14 GMT

I'm a same column fan, but with spacing.

I personally would prefer -

for (;
    {
          //foo
    }

And even if it is a single line, its quite disconcerting when people use -

if (foo)
    { //bar }
else if (foo)
    {
    //bar1
    //bar2
    }

That is so wrong - if I ever saw someone do that, I'd personally go down and smack them. I think consistency in coding is a good thing.

Coffeehouse - Why is C# Case Specific?

Karthik — Mon, 12 Apr 2004 11:54:45 GMT

SO cAn i TyPe NaTuRaL LaNGuagE lIKe ThIs tOo?

There is reason there is case sensitivity, even in written and spoken language. Maybe you should take a class of German 101.

When you are ready to capitalize in your written language, what is the problem when it comes to programming languages?

Virtually all your arguments can be used against natural language too.

I will not even get into the rest of the stuff you've written, classic Slashdot troll material.

By the way, what are you smoking? Can I have some of it too?

Coffeehouse - Any C++ guys out there?

Karthik — Mon, 12 Apr 2004 08:23:16 GMT

I agree with you for the most part.

I think that despite the development happening in C# and the like, I would still like to program in C and C++. Microsoft should stick to development of that too, and I for one would be quite bummed if MFC and VC++ were to go the way of the dodo.

Generics are cool, but I would like my templates too. Managed code is cooler, but please let me have unmanaged code when I want to.

(Although some might argue that the problems with templates, exception handling and the like are better handled in C#, but thats a whole different ball-game)

I do not want to make this sound like, BSD is dying! - but with everyone and their brother pushing Java and C#, it does sound as if they are trying very hard to convince us developers that C and C++ are dying.

I remember reading somewhere on the MS Forums that Longhorn internals have in fact been coded in C# - is that supposed to mean something?

Coffeehouse - Why is C# Case Specific?

Karthik — Mon, 12 Apr 2004 07:19:38 GMT

Richie wrote:

An "A" is not equal to an "a"... deal with it...

(I wonder what advice Donald Trump have for them...)

I wonder what advice Larry Wall would have for them

Coffeehouse - Channel 9 is great ... but needs a search engine already !

Karthik — Mon, 12 Apr 2004 06:59:52 GMT

I, I think.

If not a search feature, they could atleast use some kind of neat categorization to look at all the articles and/or posts.

But then, hey! Its just been a week or so, and am guessing that this site is going to be quite different in say, a month from now.

So far its been quite good, so lets see how this turns out. Hell, I find spending as much time here as I do on Slashdot!

Coffeehouse - Known Issues with Channel 9

Karthik — Mon, 12 Apr 2004 06:44:26 GMT

Came across a couple of other (visual) problems -

1. There seems to be a limit for usernames on the Who's Online panel on the left which causes their names to overflow out of the table (screenshot1) (screenshot2)

2. For some articles, the main panel on the right seems to go really wide (this seems to happen randomly, I've not been able to find any correlation between content and this - or maybe I've not looked hard enough (screenshot)

Cheers.

Coffeehouse - Macheads and Linux folk on Channel 9

Karthik — Mon, 12 Apr 2004 05:10:41 GMT

I use IE on Windows when I'm on wireless on my notebook - its my primary personal pc and it dual boots (XP Pro and Debian).

My work involves the use of VC++ and is for the Windows platform, so during that time I'm on IE/Mozilla (depending on what I do).

When its school stuff, the programming is mostly done in either C, Java or Perl, and so I prefer Linux.

If its personal stuff, it really depends on what is it that I want to do. Quite honestly, I would spend about 60% of my time on Windows and about 35% on Linux. I'm really uncomfortable with Macs (use them because I have to - and only commandline plus matlab) and they perhaps contribute to 5% of my use.

Coffeehouse - Ease of use

Karthik — Mon, 12 Apr 2004 04:52:54 GMT

I agree with you mate - this is true for the techie stuff too. I'm a developer and I find it difficult in finding my way across MSDN.

In fact, I always ask a friend or coworker about any problems that I have with code - it's easier than trying to find it out using MSDN. If that does not work out, I try Google.

MSDN is great - its just too organized for practical use. Its been built purely as reference - but that makes it impossible to search for specific problems. At the risk of losing my geekdom, I would say its even intimidating!

Here is a suggestion - maybe you can make MSDN into this big online KB and have a simple expert-system like front end to search for things.

I think MSDN could use a good revamp or two

Coffeehouse - Why is C# Case Specific?

Karthik — Mon, 12 Apr 2004 02:52:47 GMT

Manip wrote:

I wish wish wish the syntax checker auto-adjusted if they are typed wrong, you can even have it turned off by default to keep all the C people happy.. I dream of a day when you type "Messagebox" and it gets corrected to "MessageBox" without even bothering you. You could even take a note from MS Word's book, and if you correct it, then it doesn't try to re-correct it.

I would be scared if my IDE did that What if Messagebox was a function that I had, or was a variable? Its for the same reason why I think clippy is a pain - please let me do what I'm doing, I will take care of my mistakes myself, thank you. I design UI prototypes in VB and this is one of the many things that really really put me off.

I would have to agree with Paul Campbell that maybe it's because old habits die hard.

The point is that there are a lot of programmers out there who do not use the default IDE (I personally use emacs for all my programming - and my templates have been made to be case sensitive - my .emacs has my OO browser configured to tell me the list of variables based on my conventions - and this works well across C, C++, Java and the like). I know tonnes of other programmers who use such conventions and are used to it - and the reason they use it is primarily because its useful.

Manip wrote:

Again nobody has offered a good reason for this. Some have said easier to read bluh bluh, but someone as I would have pointed out told them automatic case selection is EASIER than human any day. I mean who has ever had problems reading Visual Basic code? Nobody!

What do you mean by automatic case selection? I like the freedom to use xfoo, Xfoo and XFoo in the same line - VB takes away that freedom and hence constrains me.

When I have 10,000 lines of code and would like to see all my local variables, conventions come in handy. Hell, even MFC follows such conventions.

Paul raised the point about Hungarian Notations, so I will not go there. But I would like to mention the much beaten down fact that the need for Hungarian Notation is perhaps mitigated in a purely OO language - but I do not think C# can be called that - I would put it together with C++ as a hybrid (a better one, but a hybrid neverthless) and therefore, there exists the need for a suitable notation.

Manip wrote:

I do suspect that because the big rubber stamp has been used on the C# specification, it is too late to change anything.. I would rant and rave about how C/C++ people control all language development but you all know what I'm going to say so I don't see the point.

Thank God for small mercies

Coffeehouse - Should C# lean more towards OO or functional languages?

Karthik — Mon, 12 Apr 2004 02:43:33 GMT

Ever so often, people argue that C# is a purely OO language, just like Java.

But I would like to say that both C# and Java lean towards being OO, and are not entirely so.

For instance, if you take Smalltalk, everything is an object. Messages are objects. All methods are incorporated into objects, and objects communicate with one another using message passing.

I will not buy the argument that OO languages are more productive than Functional Languages - I think both have their own set of benefits. The reason why C# and Java are good is only because they are managed and come with better resource management (and other resources in the form of good libraries).

So, as a C and C++ programmer, I feel the need to ask if MS leaned towards being OO so that they could compare well against Java?

Joe Beda had mentioned that managed code will be the norm, and not otherwise - and I can see the benefits of that, and even agree that it would be the right way to go. But does being managed equate to being OO? I maybe (and hope I'm) wrong, but I have a feeling MS perceives OO as being managed.

But I can see quite a lot of benefits of not going entirely OO, and maintaining functional programming attributes.

I feel that maybe MS should incorporate more functional programming features into C#, while aggresively pushing managed code.

What do you folks think?

Coffeehouse - Research at Microsoft

Karthik — Mon, 12 Apr 2004 01:47:54 GMT

paulcam wrote:

Some things do go live...

check out WWMX...

Wow, that is quite cool.

Didn't MSR also have something along these lines to catalogue a person's entire life's events? An e-database of sorts?

Coming back, I just came across this at the WWMX FAQ:

We would love to offer access control and buddy lists to the WWMX, but unfortunately, the complexities of access control and tight security prohibit a tiny team like us from implementing a satisfactory solution.

Now that is one of the reasons Microsoft could make some of their products OpenSource. I'm not suggesting all - but atleast something like this - I can think of a million things I could do if the code were to be opened up

The thing is that Microsoft does a lot of cool things with technology, but as a geek, I would like to get my hands on them and play around with atleast some of them - and maybe even contribute to it and make it better.

Mind you, I'm not even an OpenSource evangelist - I just believe in using the right tool for the right job, and understand the need for both commercial and opensource software enterprises to exist.

Its just that atleast when the work is not mainstream, and is cool enough - it makes sense to let it out into the open and let others have a go at it. Especially the work at MSR

Just my $0.02 - I hope you can see where I'm coming from.

Coffeehouse - Open Source

Karthik — Mon, 12 Apr 2004 01:36:18 GMT

I was waiting for something like this to come up

I agree with you completely. At some point of time or the other, MS will have to give a long hard look at releasing atleast some of their source into the open - there is no denying the benefits.

I wish MS would understand that both Opensource and commercial enterprises can co-exist and cater to different niches - both have their benefits and disadvantages, and there is no reason why a company cannot have part of its code proprietary and part of it closed source.

I mentioned it in this article, too. Microsoft can actually benefit by releasing some of their code out in the open - and its even been done before.

Opensource also means more eyes looking at the code, and adding extensions that could be beneficial. If not anything, MS could perhaps give this a try and see how it works out.

Coffeehouse - Why is C# Case Specific?

Karthik — Mon, 12 Apr 2004 00:51:45 GMT

ktegels wrote:

As to the back-up: no, being case insensitive is actually harder than not being. Our friends at the U of Manchester have a basic compiler theory paper up that talks about this.

Ah, thank you for the link - it was quite useful.

I said maybe because if one were to convert everything into uppercase and analyze, it would create for a smaller space to check for errors - and hence, maybe a little faster. Thats the same reason they mention why Pascal facilitates locating variables and procedures in a case-independent way - I imagined that it would narrow the possible number of comparisons, and hence maybe make make the compilation happen faster.

However, I missed the fact that token-checking and other things may use case-sensitivity as a factor - my bad.

Coffeehouse - Why is C# Case Specific?

Karthik — Mon, 12 Apr 2004 00:28:59 GMT

I think you mean case-sensitive, but I'll go ahead and tell you why I think case sensitivity is a good thing -

1. Easier to declare case specific variables, and hence results in better code - people have conventions for declaration that is done better using a case-sensitive language. xFoo, Xfoo, xfoo all can mean differently based on your conventions - I use xfoo locally within a class, Xfoo in inherited classes and XFoo for externs - its something that does not break and yet makes it obvious to me by just looking at it. And I make sure my conventions are duly noted in the documentation, and it makes it easier for others to work on it.

2. Code becomes easier to read - sure, I can write class foo as cLaSs FoO - but it looks stupid. Maintainability and readability are issues.

Manip wrote:

Doesn't this just make using the language a pain in the * and generally push up the time and complexity of development?

Do you have any basis for these claims? In fact, good coding conventions are useful in the long run.

Manip wrote:

Doesn't it make for a MORE steep learning curve?

Why? Because it does not encourage laziness and actually makes sure a programmer looks at his code and declarations, and encourages him to follow a standard convention?

I think thats a good thing. And it discourages bad programmers - which is even better.

Manip wrote:

Doesn't it make developing a syntax checking tool more time consuming?

Maybe, a little. But do you have any information to back this up?

Coffeehouse - Private Messages

Karthik — Sun, 11 Apr 2004 23:27:25 GMT

Shawn wrote:

That being said, the forum categories are pretty broad. Is there any thought being given to breaking them down any further? It doesn't need to be set in stone but it would make it a lot easier to follow along with if threads were grouped by topic.

I agree. There could perhaps be some kind of separation of the important categories that the various threads fall under?

Another suggestion would be to have some kind of message system (a la Slashdot) so that users can know when someone has replied to their posts.

(I'm aware of RSS feed, but do remember that I would access the site from a place where I may have nothing more than webaccess).

You could perhaps even have private reply-to - where only the person the reply is directed towards can read whats posted.

Coffeehouse - Research at Microsoft

Karthik — Sun, 11 Apr 2004 23:15:56 GMT

And oh - I do have to mention that Microsoft has released stuff from MSR into the open before.

For example, the MSR Downloads section has the code for some implementations (like the Context-Free Parsing Algorithms) and the like.

It would be cool if you could release more of such into the open for people to have a look at them, and perhaps help contribute.

Just a thought.

Coffeehouse - Research at Microsoft

Karthik — Sun, 11 Apr 2004 22:43:00 GMT

jonathanh wrote:

What we really need is Kevin Schofield here to talk about the process of making the jump... or maybe I should just write a book.

Think you guys could do a video on the work at MSR and how the stuff comes together? Would be really cool.

And oh, I found your Java Optimization page quite useful, thank you.

Coffeehouse - Business Week - How Microsoft is Clipping Longhorn

Karthik — Sun, 11 Apr 2004 06:58:57 GMT

Hmmm, managed code is something thats written within a managed space - think of managed space as being the front to the stuff that does all the actual work.

That way, you do not real talk to the bare-metal beneath - you just talk to a front (to quote Joel Spolsky). The point is that Microsoft is trying to achieve as close to bare-metal performance as possible while allowing for a middle-layer "safe" front to exist (managed space - in here, your threads, memory, exceptions and everything else is handled to prevent errors and overflows).

The point is that, managed code will be far less prone to the various problems that is seen in traditional "unmanaged" code.

I know where you are coming from about Java - I've been/am there right now - and its quite painful. But to be fair, Java has done a reasonably good job of it, and I guess .Net would do an equally good job (minus the perks/hassles of Java). If one were to go with the supposed benchmarks, it would look like they are getting closer to metal performance than Java does, but that could be because of lack of overheads of the JVM and the like.

And I wanted to make one another clarification - you sound as though managed code is something that MS came with - this is not so Although of supposedly dubious origin, you might want to check this.

And you might want to read Joe Beda's responses on managed code in Avalon.

Coffeehouse - When will certain Windows settings be set to more efficent settings by default?

Karthik — Sat, 10 Apr 2004 20:47:49 GMT

If you want performance boost, one thing that will definitely speed up your system is if you disabled all those graphical effects

On a serious note - all those transition effects and active desktop settings - those are the things that when disabled will speed up your system.

Your information on DLLs is wrong - by default any DLL remains mapped but not loaded in memory, and gets removed after a while (I think time-out and/or resource request). The delay in unloading is for a purpose - to help applications save their settings without killing themselves pre-emptively.

All those settings are there for a reason. By changing them, you risk losing data and affecting application performance. And then, when your applications crash more often you would be blaming Windows for it

And btw - have you looked at Windows2003? The default settings are quite robust and well done, to the point of being really BSD-ish, if I may say so.

Coffeehouse - Research at Microsoft

Karthik — Sat, 10 Apr 2004 03:58:25 GMT

If one looks at any of the major conferences related to NLP, KDDM and the like, Microsoft Research has tonnes of high quality publications that come up.

But I wonder where Microsoft goes with all this stuff - where does Microsoft utilize all that good NLP research thats being done?

I look at the projects over at MSR and I've noticed a lot of good stuff that have remained there for a while, but do not really notice them coming into the mainstream at all.

Is MSR a black-box where so many cool things happen but only some come out, or is it like a la-Bell Labs of the days of yore? Or is it Microsoft's way of IP gathering?

Personally, I would love it if MSR released some of the stuff out into the open for us geeks to play around with - hey, we may actually come up with something cool.

What do you say, folks?

Coffeehouse - Roll call - tell us who you are!

Karthik — Sat, 10 Apr 2004 00:40:20 GMT

I'm Karthik Narayanaswami, a graduate student doing my masters in HCI (CS) at Georgia Tech. I graduated last year with an engineering degree in ECE from the University of Madras.

I worked during all four years of my UG as a Research Intern on a lot of cool stuff - especially AI, HCI, Graphics and Software Engineering. Right now, I'm working mostly on the use of my EE/AI/Graphics skills in HCI in creating anthropomorphic intelligent agents and robots.

I'm a CS, EE, Physics and Music geek. Other non-geeky things I do are rock-climbing, live concerts (the rock variety), travelling and reading.

And at the moment, I'm hunting for internships and even have an interview scheduled at Redmond sometime this month end!

Coffeehouse - Known Issues with Channel 9

Karthik — Fri, 09 Apr 2004 20:31:03 GMT

Just a UI suggestion - when you hit tab, its usually assumed that it would take the user to the post/submit button or at the very least outside the input text-box.

For example, I just instinctively tend to hit a couple of tabs to go to the submit button on almost all forums. But everytime I try doing that here, I realize that tab is just for within the rich text-box.

I've noticed that Hotmail has the same problem - initially, it was designed such that a couple of tabs would take you to the send button, but now it *needs* to be clicked.

It breaks existing functionality, and I just feel that its bad HCI

My $0.02