I've had a Dialog Mobile Broadband (HSPA) connection for an year now. I don't have much to complain about its connectivity issues like most of the others do. But definitely I'm not satisfied with their service....It's too costly for what they offer.
They have a very unreal fair usage policy - After the downloads exceed 5GB, the speed goes down to 384kbps and then after 6GB....it becomes 64kbps! That's totally useless stone-age connection and it always remembers me how I went online in 2002 with a dial-up modem. What can you accomplish with a 64kbps dial-up like connection?
I waited for some time thinking that they will change their FUP to be more practical, but it never seemed to happen. Now that the time has come for a change and today I bought an Airtel HSPA connection. Here are the speeds they promise they are offering:
Downloads upto 10 GB -> 1.8 Mbps downlink (384 kbps uplink)
10 GB - 40 GB -> 512 kbps downlink
40 GB onwards -> 256 kbps downlink
This looks to be a much more usable connection compared to the Dialog's one. Both connections have the same tariff rates. I haven't tested the connection yet.... But hope this will produce a better Mobile Broadband experience.
Tuesday, April 6, 2010
Monday, March 15, 2010
Simple stuff missing in .NET and Visual Studio
I'm a Java programmer, but today I had to deal with C#.NET to create a simple Windows Forms application which had some mathematical and algorithmic stuff.
I've heard a lot of positive comments about Visual Studio, but what I experienced today was not up to the level of usual praises it gets from a lot of developers. Compared to IntelliJ IDEA, which is my favorite IDE, there are several stuff missing in Visual Studio (2008 edition). First thing was auto-complete. Of course it does have a reasonable auto-complete facility, but it's partial (at least with the default settings - I don't know whether there's a way to change it). Whenever a method name is completed, we have to manually code '(' and ')' parts of the method call, which I found a bit inconvenient. The same seems to be the case with strings, once I open a double-quote, the IDE doesn't automatically close it - it's inconvenient when there's a lot of such coding to be done.
Second noticeable thing was that most often it doesn't automatically point out the coding errors unless we build it manually, this is troublesome when there's a lot of coding to be done. Then there was no way of altering editor's background color - it's set to bright white by default and can be really troublesome for the eyes in the long run.
Finally, the most weird part, I wanted to use a Hashtable to have access to some values using keys. I defined one and then wrote code to add a lot of stuff. Then when I was looking for a method to retrieve them based on keys......there was no such simple method retrieve elements like in Java's Hashtables. (there was a method to check whether an item exists, but can't fetch it directly - how terrible is it?) After a Google search, I found a lot of articles teaching how to fetch an item in C# Hashtable - iterate through all elements....! This is really useless and a terrible hashtable implementation I've ever seen. Of course they might have written efficient methods to put stuff (I assume at least those Add methods are efficient), but there was no efficient or an easy way of retrieving them back. For algorithms which have a lot of get calls rather than inserting elements, this will cause unnecessary overhead.
Although I have no idea of moving to .NET, hope these will be fixed in upcoming Visual Studio 2010 edition.
I've heard a lot of positive comments about Visual Studio, but what I experienced today was not up to the level of usual praises it gets from a lot of developers. Compared to IntelliJ IDEA, which is my favorite IDE, there are several stuff missing in Visual Studio (2008 edition). First thing was auto-complete. Of course it does have a reasonable auto-complete facility, but it's partial (at least with the default settings - I don't know whether there's a way to change it). Whenever a method name is completed, we have to manually code '(' and ')' parts of the method call, which I found a bit inconvenient. The same seems to be the case with strings, once I open a double-quote, the IDE doesn't automatically close it - it's inconvenient when there's a lot of such coding to be done.
Second noticeable thing was that most often it doesn't automatically point out the coding errors unless we build it manually, this is troublesome when there's a lot of coding to be done. Then there was no way of altering editor's background color - it's set to bright white by default and can be really troublesome for the eyes in the long run.
Finally, the most weird part, I wanted to use a Hashtable to have access to some values using keys. I defined one and then wrote code to add a lot of stuff. Then when I was looking for a method to retrieve them based on keys......there was no such simple method retrieve elements like in Java's Hashtables. (there was a method to check whether an item exists, but can't fetch it directly - how terrible is it?) After a Google search, I found a lot of articles teaching how to fetch an item in C# Hashtable - iterate through all elements....! This is really useless and a terrible hashtable implementation I've ever seen. Of course they might have written efficient methods to put stuff (I assume at least those Add methods are efficient), but there was no efficient or an easy way of retrieving them back. For algorithms which have a lot of get calls rather than inserting elements, this will cause unnecessary overhead.
Although I have no idea of moving to .NET, hope these will be fixed in upcoming Visual Studio 2010 edition.
Sunday, February 14, 2010
Is 'private' really 'private'?
I've got the following Java class with private methods.
class ClassX {
private void methodX() {
System.out.println("methodX: this is private.");
}
private void methodY() {
System.out.println("methodY: this is private.");
}
}
How to invoke these methods from an outside class without modifying this code?
If you've studied Java, you might have learnt that it's impossible.
Truth is, you can access anything from anywhere.
Just use Reflection.
class Main {
public static void main(String[] args) throws Exception {
ClassX cx = new ClassX();
Method method = ClassX.class.getDeclaredMethod("methodX", null);
method.setAccessible(true);
method.invoke(cx, null);
}
}
All you have to do is to take the method using 'getDeclaredMethod', then set it as 'accessible' (previously it was not accessible since it was private). Now you can do any damn thing with it. How simple is it?
Still not enough? You can iterate through all declared methods in a class. (in case it is a compiled one and you don't already know what private methods are there)
Method[] ms = ClassX.class.getDeclaredMethods();
for (Method m: ms) {
System.out.println(m.toString()); // see what it is.
m.setAccessible(true);
m.invoke(cx, null); // invoking each method.
}
You can do the same thing for variables as well.
Is 'private' really 'private'? No.
Then how to make something really 'private'? Who knows?
That's not the important point here. If you don't know about reflection, then go ahead and learn it. Reflection is what enables real 'dependency injection' in Java.
class ClassX {
private void methodX() {
System.out.println("methodX: this is private.");
}
private void methodY() {
System.out.println("methodY: this is private.");
}
}
How to invoke these methods from an outside class without modifying this code?
If you've studied Java, you might have learnt that it's impossible.
Truth is, you can access anything from anywhere.
Just use Reflection.
class Main {
public static void main(String[] args) throws Exception {
ClassX cx = new ClassX();
Method method = ClassX.class.getDeclaredMethod("methodX", null);
method.setAccessible(true);
method.invoke(cx, null);
}
}
All you have to do is to take the method using 'getDeclaredMethod', then set it as 'accessible' (previously it was not accessible since it was private). Now you can do any damn thing with it. How simple is it?
Still not enough? You can iterate through all declared methods in a class. (in case it is a compiled one and you don't already know what private methods are there)
Method[] ms = ClassX.class.getDeclaredMethods();
for (Method m: ms) {
System.out.println(m.toString()); // see what it is.
m.setAccessible(true);
m.invoke(cx, null); // invoking each method.
}
You can do the same thing for variables as well.
Is 'private' really 'private'? No.
Then how to make something really 'private'? Who knows?
That's not the important point here. If you don't know about reflection, then go ahead and learn it. Reflection is what enables real 'dependency injection' in Java.
Sunday, December 20, 2009
Total Freedom
Recently I accidentally found a completely new software license called WTFPL. It is a license which is dedicated to deliver true freedom and to make the lives of the people easier.
Initially I thought this was something created just for fun, but that was not the case. I found several real open source projects licensed under WTFPL. :D
At a glance, it appeared to me as this license should be compatible with most of the other licenses as it allows you to do whatever you want (I'm avoiding the original way used by the license creator to say this :P).
When I first read GPL and BSD license statements, I couldn't understand anything. I had to find articles written in simple language which explain what the licenses were all about. But this is much more simpler. Obviously, this is the first license I understood at the first glance. :P
So if you're concerned about giving total freedom to the users/developers, then WTFPL is something you should consider when you decide a license for your project. Seriously.
Initially I thought this was something created just for fun, but that was not the case. I found several real open source projects licensed under WTFPL. :D
At a glance, it appeared to me as this license should be compatible with most of the other licenses as it allows you to do whatever you want (I'm avoiding the original way used by the license creator to say this :P).
When I first read GPL and BSD license statements, I couldn't understand anything. I had to find articles written in simple language which explain what the licenses were all about. But this is much more simpler. Obviously, this is the first license I understood at the first glance. :P
So if you're concerned about giving total freedom to the users/developers, then WTFPL is something you should consider when you decide a license for your project. Seriously.
Wednesday, November 25, 2009
Microsoft Fonts for Firefox on Ubuntu...
When I first started using Ubuntu, I found that the default fonts come with its Firefox are a bit difficult to read. By default, Ubuntu doesn't have popular Microsoft fonts such as 'Times New Roman' pre-installed. So the first task I did was to install Microsoft True Type core fonts bundle on Ubuntu. It can be done with a simple apt-get command as following:
sudo apt-get install msttcorefonts
This installs all the required Microsoft fonts. Then all I had to do was just to change the font settings in Firefox to be similar to Firefox on Windows.
Preferences -> Content
Default Font: Times New Roman
Size: 16
In Advanced Settings:
Fonts for: Western
Propotional: Serif Size: 16
Serif: Times New Roman
Sans-serif: Arial
Monospace: Courier New Size: 13
Note that these settings are similar to Firefox on Windows 7.
That's it.
sudo apt-get install msttcorefonts
This installs all the required Microsoft fonts. Then all I had to do was just to change the font settings in Firefox to be similar to Firefox on Windows.
Preferences -> Content
Default Font: Times New Roman
Size: 16
In Advanced Settings:
Fonts for: Western
Propotional: Serif Size: 16
Serif: Times New Roman
Sans-serif: Arial
Monospace: Courier New Size: 13
Note that these settings are similar to Firefox on Windows 7.
That's it.
Thursday, October 29, 2009
Progress of epZilla
I'm involved in doing a research on a scalable and a fault-tolerant distributed architecture for complex event processing (epZilla) as my final year undergraduate project with three other colleagues. So far it has been all research and no coding, which makes life a bit harder for us coz we are not very good readers and also it makes life even harder in evaluations by the department since we don't have anything other than some research papers and some ideas and concepts in mind which doesn't seem to be very much convincing for the evaluators.
So far we have read several dozens of research papers on many related areas and now about to finalize the architecture. Meanwhile we are about to start some coding even before we finalize the architecture since we gotta have something working before the next evaluation - for demonstration purposes.
I'm posting this here so that I can look back later and find how our project evolved.
So far we have read several dozens of research papers on many related areas and now about to finalize the architecture. Meanwhile we are about to start some coding even before we finalize the architecture since we gotta have something working before the next evaluation - for demonstration purposes.
I'm posting this here so that I can look back later and find how our project evolved.
Monday, September 21, 2009
Ideas from Code Jam....
Recently, I accidentally got to know about Google's Code Jam competition. After getting to know that it is an algorithmic problem solving competition, I decided to give it a try. I've never had much experience in this area, so I did not have any hope of advancing to the semi-finals or finals but thought that this is a good opportunity to sharpen thinking ability.
I got through the qualification round with ease as it contained two problems which could be solved without knowing any advanced problem solving technique. But for the next round, it clearly appeared like knowing only to solve number theory, brute-forcing/backtracking and ad-hoc problems wouldn't help. So I Googled a bit to find what to learn.......and found several problem type classifications which were based on different programming competitions. Here's Hal Burch's classification based on ICPC which is somewhat applicable to Code Jam too:
Of course, many of these types can be solved using simple brute-force algorithms, but they will end up taking a few million years to give you an answer for some of the inputs..... so definitely you need to learn some efficient and easy to implement techniques to tackle such problems. The best resource I found for this is TopCoder's algorithm tutorials, which contain no-nonsense and easy to understand tutorials on most of these areas.
Having said that, I should now come to the point.......that is, some of the smart tricks used by the competitors in this Code Jam competition. One is that, since the solutions are executed in competitor's mahine itself, it gives competitors the ability to generate all solutions to a problem using brute-force which might take a couple of hours (before downloading the input, coz once you download inputs you'll have only 8 minutes to solve them), save it, then download the actual input and then just lookup the already generated results to directly find answers to the input. No DP, no efficient algorithms, but still perfectly legal and effective! (if your computer has enough processing power). I actually found a few guys who had advanced to round 2 using this technique.
Second one is writing multi-threaded programs. It feels like a crazy idea trying to implement a multi-threaded algorithm.....but think about it.....you might have a Core 2 Duo processor, if you write a normal inefficient algorithm, it will execute on only one core.....and your program might take 8 minutes plus a couple more seconds to execute and you'll be out of the competiton. What a waste.....you might have implemented the same program to have multiple threads and execute over both cores and could have solved it within 8 minutes! (this is a rare case, but actually happened in Round 1A - Problem A) Now some of the competitors are interested in writing toolkits to provide concurrent execution over all cores they have on all computers at home, be it a desktop computer, a laptop or a netbook. What a fantastic idea!....they've turned this competition into a concurrent programming competition too. By the way, I have only one single-core machine at home, so I'm trying to find an online supercomputer service which can offer some computing time for me to execute my horrible brute-force programs to advance to the next round :) Just another different idea ;)
Anyway, the competition is getting fierce....so there's a very little chance (most probably no chance at all) to advance through round 2 (which takes only top 500 contestants). But if you too give it a try, you'll certainly discover some fantastic techniques to tackle problems.
I got through the qualification round with ease as it contained two problems which could be solved without knowing any advanced problem solving technique. But for the next round, it clearly appeared like knowing only to solve number theory, brute-forcing/backtracking and ad-hoc problems wouldn't help. So I Googled a bit to find what to learn.......and found several problem type classifications which were based on different programming competitions. Here's Hal Burch's classification based on ICPC which is somewhat applicable to Code Jam too:
- Dynamic Programming
- Greedy
- Complete Search
- Flood Fill
- Shortest Path
- Recursive Search Techniques
- Minimum Spanning Tree
- Knapsack
- Computational Geometry
- Network Flow
- Eulerian Path
- Two-Dimensional Convex Hull
- BigNums
- Heuristic Search
- Approximate Search
- Ad Hoc Problems
Of course, many of these types can be solved using simple brute-force algorithms, but they will end up taking a few million years to give you an answer for some of the inputs..... so definitely you need to learn some efficient and easy to implement techniques to tackle such problems. The best resource I found for this is TopCoder's algorithm tutorials, which contain no-nonsense and easy to understand tutorials on most of these areas.
Having said that, I should now come to the point.......that is, some of the smart tricks used by the competitors in this Code Jam competition. One is that, since the solutions are executed in competitor's mahine itself, it gives competitors the ability to generate all solutions to a problem using brute-force which might take a couple of hours (before downloading the input, coz once you download inputs you'll have only 8 minutes to solve them), save it, then download the actual input and then just lookup the already generated results to directly find answers to the input. No DP, no efficient algorithms, but still perfectly legal and effective! (if your computer has enough processing power). I actually found a few guys who had advanced to round 2 using this technique.
Second one is writing multi-threaded programs. It feels like a crazy idea trying to implement a multi-threaded algorithm.....but think about it.....you might have a Core 2 Duo processor, if you write a normal inefficient algorithm, it will execute on only one core.....and your program might take 8 minutes plus a couple more seconds to execute and you'll be out of the competiton. What a waste.....you might have implemented the same program to have multiple threads and execute over both cores and could have solved it within 8 minutes! (this is a rare case, but actually happened in Round 1A - Problem A) Now some of the competitors are interested in writing toolkits to provide concurrent execution over all cores they have on all computers at home, be it a desktop computer, a laptop or a netbook. What a fantastic idea!....they've turned this competition into a concurrent programming competition too. By the way, I have only one single-core machine at home, so I'm trying to find an online supercomputer service which can offer some computing time for me to execute my horrible brute-force programs to advance to the next round :) Just another different idea ;)
Anyway, the competition is getting fierce....so there's a very little chance (most probably no chance at all) to advance through round 2 (which takes only top 500 contestants). But if you too give it a try, you'll certainly discover some fantastic techniques to tackle problems.
Subscribe to:
Posts (Atom)