Channel 9 - Discussions by SvendTofte

Coffeehouse - LaTeX anyone?

SvendTofte — Tue, 27 Sep 2005 08:41:22 GMT

I do. LaTeX produces beautiful math IMO.

Coffeehouse - Can someone explain video to me?

SvendTofte — Sat, 24 Sep 2005 00:36:03 GMT

In the past, when QT "was it", for the net (and IMO still is), it had the Sorenson codec, which was QT only, and was pretty sharp about compressing. This was a long time ago though.

Coffeehouse - Mozilla is a virus

SvendTofte — Wed, 21 Sep 2005 18:11:44 GMT

This site is to MS as Slashdot is to Linux. Did you actually try and read the link you posted? Is MS responsible if idiots downloads patches from p2p, and find that they get infected? The Korean server is not an affiliated Mozilla site, it's just any old fan site. Yawn.

Coffeehouse - Asymptotic Analisys

SvendTofte — Tue, 13 Sep 2005 12:34:46 GMT

amotif wrote:

Shmuelpro wrote: oh common can anyone at least tell me that they know what it is

I think we most often refer to "Big-O notation" in discussing the magnitude of algorithms... is that where you're going with this?

There's several versions of algoritihmic analysis. Big O is the most common, and often it's quite intuitive. IIRC, O is the maximum time/resource that will ever be spent. You have other symbols for things, like "minimum possible resource expenditure", and average. Average runtime analysis is hard, but needed.

Quicksort is a O(n^2) algorithm. Like Bubble sort. That is, the runtime of the algorithm is proportional to the square of the length of the input (the number of items to be sorted).

If you do propability analysis on the formular however, you will find the run time to near O(n log n), which is a theoretical lower bound for comparison based sorted (bucket sort is in theory faste, as it does not compare, in order to sort). QS will only ever hit the worst time, when it's splitting the array to be sorted, when selection of the pivot point results in a constant split, such as:

[2 elem] [pivot elem] [rest of elem]

However, bad luck like this is extremely unlikely, and as long as you're getting splits, like this:

[x % of elems] [pivot elem] [rest of elem]

Then the time it takes, will be O(n log n).

I don't know what Wikipedia says on the topic, but the page on QS is awefully long: http://en.wikipedia.org/wiki/Quicksort

A pretty standard book on the analysis of algorithms and datastructures is the "Introduction to Algorithms" from MIT Press, by Cormen, et al.

Coffeehouse - Whats an EXE?

SvendTofte — Sat, 10 Sep 2005 19:18:02 GMT

start -> run -> cmd -> debug -> go nuts!

Tech Off - Regular Expressions

SvendTofte — Sat, 10 Sep 2005 16:12:52 GMT

I'll assume, that you mean validating string versions, of these datatypes? Otherwise, regular expressions will not do much. That said, you'll probably want to look at the specifications for each language. I don't know about VB.NET, but you can find C#'s here: http://msdn.microsoft.com/library/default.asp?url=/library/en-us/csspec/html/vclrfcsharpspec_c.asp You'll notice that there only exists an "integer-literal" and "real-literal". Exactly what this is taken as, I'll assume depends on what it's either prefixed, or postfixed with, or what the context is. As such, there are no explicit regular expressions (or even implicit, the definition is grammatical, not lexical).

Coffeehouse - windows "genuine" validation

SvendTofte — Fri, 10 Jun 2005 12:42:44 GMT

l33tn00b wrote:

i know its an unpopular viewpoint, but i think validaiton should be required for all downloads from microsoft.com . people who are honest and do not pirate should be rewarded.

Not everyone who has a pirated Windows copy actually pirated it themselves.

I don't know how many percent this is, but it's probably a good bunch of percent.

Coffeehouse - Discussion: Is CSS layout overrated?

SvendTofte — Mon, 30 May 2005 12:15:09 GMT

mVPstar wrote:

The beauty of a CSS layout is the fact that in can degrade nicely on clients that don't support CSS.

This is one fallacy I'm pretty tired of seeing. While it's literally true, A website written using pure CSS and HTML will usually degrade well to a totally CSS-oblivious client.

The problem is that there are SO FEW totally CSS-oblivious clients. We don't care about bots and stuff, so that leaves us with small devices pretty much, and perhaps some specialty browsing device for handicapped (though most visually impaired use IE for browsing).

The problem is that by far the largest number of clients (like in the 99% range) have varied and mixed support for various parts of the CSS standard.Imagine a browser supporting background-image, but not background-repeat (NS4!). It could potentially render the page unreadable. This whole issue is even why we spend so long fighting IE, FF, or Opera, or whatever browser is having a problem, because CSS in effect does no degrade gracefully. And if we don't check it in said browser, the page can literally become unreadable.

That said, CSS is a heck of alot better then the alternative choice. Years back, I've nested tables (on commercial sites) up to nine levels, and then it was all framed even! And the cascade is simply a beautiful thing.

But people also should see that CSS is only meant to go so far. CSS in my opinion scales like (I need to watch my language), and should not be used, at all, for say behaviour, apart from the most simplistic :hover effect. Building entire menus out of CSS rubs me the wrong way...

The Sandbox - Colour Filter

SvendTofte — Sat, 21 May 2005 17:50:32 GMT

it must be immensely slow if it's using GetPixel. Rewrite to it is using pointers (assuming C# here)

Coffeehouse - What Keyboards do you use ?

SvendTofte — Fri, 13 May 2005 15:01:25 GMT

Wow, a forum full of programmars, and no followers of the IBM Model M keyboard??

Coffeehouse - More Sun interoperability

SvendTofte — Tue, 10 May 2005 14:00:45 GMT

Sure there are, and MS's combined efforts are showcased here:

http://browsers.evolt.org/?ie/solaris

(I have never tried IE for Unix, but I have heard that it blows, besides being eons old of course)

.
..
...

What desktop users even use solaris? surely, only desktop users are really interested in IE, wmp and Office ... (if you read joel, you know who uses Solaris btw.)

I have no idea about the installed bases of Solaris, the users on it, or Microsoft's ability to port WMP/IE/Office readily to other platforms (Solaris might not even be x86), but this just strikes me as ... so out there. What would MS gain?

If anything, it would only make sense to port SQL Server, and things like that. At least people running stuff like that on Solars are willing to shell out big bucks.

Tech Off - Question regarding AJAX

SvendTofte — Sun, 08 May 2005 18:00:48 GMT

"AJAX" is only asynchronous on the client.

If you want to make it async on the server too, look at how to push code. You can, forinstance "stream" a JavaScript file, by continually feeding empty spaces to the client, and, when you wish, stream a JS statement.

Coffeehouse - Ross... What is your avatar?

SvendTofte — Thu, 28 Apr 2005 05:33:20 GMT

http://www.toyvault.com/cthulhu/ ?

Tech Off - Who needs another programming language?

SvendTofte — Mon, 25 Apr 2005 07:18:03 GMT

My first thought is: have you ever written a (even small) language before? Small toy languages, with no idea or concept, can probably be done in a few weeks by a newbie. If it needs a (single) unqiue concept (but is still toy), then expect months. If you want fully fledged super language, well ... I'd give up there

Everyone will probably spout the "go read the dragon book", which you also should. Writing compilers is a wonderful exercise, touching on MANY aspects of computer science. Plus, you get a gold star on your geek card.

For books, some solid entry level books, are these:
http://www-db.stanford.edu/~ullman/ialc.html (I've only the first ed, which is solid)
http://www.amazon.com/exec/obidos/tg/detail/-/0201100886/002-6508715-5124031?v=glance (which I heard was coming in a 2nd ed?)

Now. Just for some loose comments, because it's nine in the morning, and I'm yet to go to bed ...

1.

It doesn't make much sense to make a language "Windows oriented", or made for "Linux". It's a language, it will run on x86, or PowerPC, or the Alpha (!). Focus instead on creating a interop mechanism, so that users of your language can leverage as much existing code (be it in the Windows API, or their own DLL's or whatever) as possible.

Edit: Why not make your language run on the .NET runtime? It seems to be all the rage these days

2.

In my ears, your wishes for "easy-to-use", and "easy to do low-level" (hardware) seems clashing. I can only think of C++, which solves this problem (it is possible to write nice looking C++ code, and then suddenly, assuming the enviroment supports it, reach through down to the machine).

Languages which has beautiful concepts (functional, ml's type system, etc), would not live well with hardware programming. It simply doesn't mesh.

I don't want to come across as elitists. But in my opinion, languages should not (and logically, cannot) do it all, for everyone. Classic example is SQL. Does some stuff well, and others not so well. But using SQL for what it was meant for, most will agree it's doing a good job.

3.

Programming is simply not easy. "public static void main" may seem obtuse at first, but after learning the basics, it's not anymore. All the keywords can be accounted, and explained away. Should it be called "starthere main" instead? Remember, that special-cases are at least one thing, which makes languages hard to read.

But anyway. Programming is not easy. It cannot be made easy, by creating a language, whose surface syntax is easily read. The programmer must still juggle concepts and abstractions in his mind. You cannot write any language, which removes this.

Of course, some languages allow a more natural approach to a problem domain, then others. But I'm not seeing that one language can give a natural approach, to all problem domains.

4.

Don't say internet.

But anyway, good luck with it all. Language design is fun and very challenging.

Coffeehouse - QuickBoard?

SvendTofte — Thu, 07 Apr 2005 01:07:21 GMT

littleguru wrote:

It would simply be a clipboard that takes more then one item. I hardly miss that in Windows too. I don't understand why they don't implement it... Office has one, but you have always to have office running to use it...

There's something nice about having just one buffer.

And this isn't like some Apple mouse argument, but rather, you can't SEE what you have in your buffer, so anything you keep in there will live as long as your shorterm memory will keep it alive (I'd guess some 30 secs max, more if you actively try to remember it ... ). Adding, say ten slots of copy/paste slots, adds a LOT of complexity to the user experience. First off, you cannot see what you have in your buffer. So you will not be able to tell what you have in the ten buffers. And which one had the text you wanted for that bit? Well, there is only one way to find out, to probe your buffers, and iteratively try your way forward. And when you want to copy something, you're faced with the additional burden that it must go into some buffer of your choice ... But which one? You don't want to erase your bank account number, since you'll need that in a while, but you don't need your moms phone number anymore ... but where did you put that? See where I'm getting at?

In the end, you must devise your own ad-hoc system, such as "buffer one is the scratch buffer, two is for ..." and so on. And really, you so rarely need this. I often use VIM, and it has 10 (or is it 26?) buffers, and I never use anything but the default buffer.

Much better to view the buffer for what is it: a volatile store for information, designed ONLY to live as long as a users short-term memory lets it.

If you consistently need to copy texts like that, I'm sure there are programs out there that will do that.

Coffeehouse - No Postcounts

SvendTofte — Mon, 28 Mar 2005 08:00:27 GMT

I only ever really think PC is a problem on immature boards. On boards with more reasonable people, it does several good things:

1. shows which is more likely to be knowlegable, especially in the "ways" of the board
2. it offers a reward, a high PC is a official stamp of authenticity. You've been around.
3. It let's the community build a sense of fellowship, letting you recognize regulars, without having to know their screen names (some communities are rather large).

As long as no further reward is given (rank, and stuff like that), people will hardly ever post-spam.

And what are these "n00b" comments? Experienced posters, calling news for idiots? Based on their PC?

I really don't think PC is anything bad. It's a small inobtrusive stat, that does not clutter anything up, plus it tells a great deal (it's not a certificate, but wanna bet there's a corrolation?).

/ST's PC:2

The Sandbox - DOS versions of: alias, cat, cp, ls, mv, rm, xargs - and desktop.bat

SvendTofte — Fri, 25 Mar 2005 13:14:03 GMT

http://unxutils.sourceforge.net will also do most. native builds for win32, so no strange cygwin dependencies. the ls that ships above doesn't do colours, however this one does: http://utools.com/msls.htm just fyi

Coffeehouse - SOA Hype

SvendTofte — Wed, 23 Mar 2005 23:11:09 GMT

I am the only one slightly amused by the fact that the editor of XML.com seems to think that SOA is hype? /new around here, hope I'm using the forum software correctly...