During a disussion on Xen-users mailing list, I found this piece of text from Eric Johansson. It is very amusing geeky humoros quote: "yes, I tried the debootstrap route and sat there watching in childlike wonder as it presented me with a rainbow of failures. Left me muttering something about rocket club open source projects as I am often given to doing lately."
On an independent issue, while just logging into blogger.com I realized that the animation on its login page causes firefox to stammer on a linux machine. I say 'stammer' because the username-password I punch gets batched and appear abruptly later.
And firefox users! Try Aquatint - a fantastic new theme!
Monday, December 12, 2005
Sunday, December 11, 2005
Ease of Linux 2.6 and Virtualization
During thanksgiving I built from the linux 2.6 source tree and realised how easy it has become to install a linux kernel from the source. 3-4 make targets and you have new kernel deployed in /boot/ with grub.conf neatly updated. Maybe because I had tough time with many experimental kernels in the past, I am finding this ease of installation surprising.
And in past 2-3 days I had a similar experience with new version of Xen - Xen 3.0. Although I heard that 'yum update' in Fedora Core 4, would install the latest Xen for me, that hadn't worked for me few months back. So I had shelved it for a while. After learning about new release of Xen, I decided to build xen 3.0. Similar to Linux 2.6, the building and installing of Xen 3.0 is very smooth. I however had first hiccup when running 'xend'. It was because of some version conflicts in some libraries, because I still had old FC4 Xen on my system. Ian Pratt precisely pointed that out on xen-users mailing list and that fixed it. Now I can run xend and tried to fork new virtual machines. I am now having problems with the file-system of my init ramdisk. Not giving ramdisk makes the VM hang at one point. I need to figure put the booting process through init ramdisk in detail to resolve this isssue.
It is amazing to know that with Intel's new VT (Vanderpool) and AMD's pacifica processors, Xen can run Windows XP unmodified. As I understand it these new hardwares will be adding a new protection level (remember those protection rings from 0 to 3 in 386 architecture?) for the VMM's hypervisor. I found it amusing that x86 architecture gave us 4 protection levels to work with and most of the OSes use just 2 of them - as for linux 0 for kernel and 3 for user programs, yet for the hypervisor technology, they needed an additional protection level outside this range. It seems that Intel VT have started shipping, but read somewhere that they are early test versions. AMD pacifica are due in early 2006. I was wondering should I wait till pacifica comes out to build my new desktop. But may be they will be early unreliable versions and probably pretty expensive. And I can't wait to play NFS most wanted, till the hardware virtualization technology stabilizes.
And yes... yesterday I tried these two firefox plugins - GSpace and Gmail notifier. GSpace finally gives a working tool to store your files on internet using your gmail account. Besides this... how can I forget del.icio.us? It has solved my long pending problem of having unified bookmarks. I use firefox/opera on windows/linux on my home/office machines. So that gives me 8 bookmark lists to synchronize. Lately I had given up synchronizing them. But with an account with del.icio.us, I can save them on internet and can access from any machine/browser/OS. They also have an import tool which I figure will allow me to import my existing bookmarks into it. No need to say, I will be spending more of my time in coming week to find new plugins around del.icio.us. :)
Xen, GSpace, Del.icio.us - When I started writing this post, I wasn't finding any common theme for this post. But now I have one - Virtualization. These three pieces of software - miles away from each other in terms of their execution domain - highlight a single trend that will reign the information technology now - 'Virtualization'. Xen virtualizes hardware from the operating systems so that multiple operating systems can run on a machine simultaneously and a single instance of operating system can be dispatched to other hardware on the fly. GSpace achieves storage virtualization by storing the files on a practically omni-present media called 'Gmail'. Del.icio.us works on the same lines by hosting our bookmark lists on some web server.
And in past 2-3 days I had a similar experience with new version of Xen - Xen 3.0. Although I heard that 'yum update' in Fedora Core 4, would install the latest Xen for me, that hadn't worked for me few months back. So I had shelved it for a while. After learning about new release of Xen, I decided to build xen 3.0. Similar to Linux 2.6, the building and installing of Xen 3.0 is very smooth. I however had first hiccup when running 'xend'. It was because of some version conflicts in some libraries, because I still had old FC4 Xen on my system. Ian Pratt precisely pointed that out on xen-users mailing list and that fixed it. Now I can run xend and tried to fork new virtual machines. I am now having problems with the file-system of my init ramdisk. Not giving ramdisk makes the VM hang at one point. I need to figure put the booting process through init ramdisk in detail to resolve this isssue.
It is amazing to know that with Intel's new VT (Vanderpool) and AMD's pacifica processors, Xen can run Windows XP unmodified. As I understand it these new hardwares will be adding a new protection level (remember those protection rings from 0 to 3 in 386 architecture?) for the VMM's hypervisor. I found it amusing that x86 architecture gave us 4 protection levels to work with and most of the OSes use just 2 of them - as for linux 0 for kernel and 3 for user programs, yet for the hypervisor technology, they needed an additional protection level outside this range. It seems that Intel VT have started shipping, but read somewhere that they are early test versions. AMD pacifica are due in early 2006. I was wondering should I wait till pacifica comes out to build my new desktop. But may be they will be early unreliable versions and probably pretty expensive. And I can't wait to play NFS most wanted, till the hardware virtualization technology stabilizes.
And yes... yesterday I tried these two firefox plugins - GSpace and Gmail notifier. GSpace finally gives a working tool to store your files on internet using your gmail account. Besides this... how can I forget del.icio.us? It has solved my long pending problem of having unified bookmarks. I use firefox/opera on windows/linux on my home/office machines. So that gives me 8 bookmark lists to synchronize. Lately I had given up synchronizing them. But with an account with del.icio.us, I can save them on internet and can access from any machine/browser/OS. They also have an import tool which I figure will allow me to import my existing bookmarks into it. No need to say, I will be spending more of my time in coming week to find new plugins around del.icio.us. :)
Xen, GSpace, Del.icio.us - When I started writing this post, I wasn't finding any common theme for this post. But now I have one - Virtualization. These three pieces of software - miles away from each other in terms of their execution domain - highlight a single trend that will reign the information technology now - 'Virtualization'. Xen virtualizes hardware from the operating systems so that multiple operating systems can run on a machine simultaneously and a single instance of operating system can be dispatched to other hardware on the fly. GSpace achieves storage virtualization by storing the files on a practically omni-present media called 'Gmail'. Del.icio.us works on the same lines by hosting our bookmark lists on some web server.
Friday, November 18, 2005
Natural keyboard
I was doing little research on natural keyboards (it's been a while that I have bought some electronic gadget) and I came across some of these amusing facts.
This particular article from Inquirer says 'natural' keyboard is a wrong design. I didn't know that shifting to natural keyboard needs some learning (which is apparently less for touch typists).
This is another funny link that tells how to make natural keyboard from regular one.
But now I have abondoned the plan of buying a new keyboard. :)
This particular article from Inquirer says 'natural' keyboard is a wrong design. I didn't know that shifting to natural keyboard needs some learning (which is apparently less for touch typists).
This is another funny link that tells how to make natural keyboard from regular one.
But now I have abondoned the plan of buying a new keyboard. :)
Friday, October 21, 2005
Inattentive blindness
A friend showed me this cool video clip.
http://viscog.beckman.uiuc.edu/grafs/demos/15.html
BUT Before watching it read this: There are two teams of people, one wearing white clothes and other wearing black clothes. Each team has a ball that they throw to another member of the same team. Your job is to find out how many throws does the white team members do.
OK, now watch the video.
Did you find anything wierd. No? Then go back and watch the video again. But only this time, don't work on counting the number of throws of the balls, just watch the video relaxed.
Did you see what you missed first time?
Cool... isn't it?
Enjoy!!!
http://viscog.beckman.uiuc.edu/grafs/demos/15.html
BUT Before watching it read this: There are two teams of people, one wearing white clothes and other wearing black clothes. Each team has a ball that they throw to another member of the same team. Your job is to find out how many throws does the white team members do.
OK, now watch the video.
Did you find anything wierd. No? Then go back and watch the video again. But only this time, don't work on counting the number of throws of the balls, just watch the video relaxed.
Did you see what you missed first time?
Cool... isn't it?
Enjoy!!!
Saturday, October 08, 2005
Do what I mean
When we hear about the user interface talks, one theme always sticks out 'Do what I mean'.
We expect our gadgets, our software programs to do what we mean and not what we tell them. But this is a challenging problem. The means by which we can "tell" our gadgets what we want, are not sufficient to convey them what we "mean". This is the gap between syntax and semantics.
The innovations in User Interface strive to bridge this gap. Ipod is an excellent example of such innovation. The advances in the desktop GUI is another.
What is it that makes it so hard a problem? Or why is this a problem at all? Here are my 2 cents on this topic, in context of my Ipod.
Why is ipod's interface more user friendly than previous music players?
Let us first see, what we want a music player to do? - play a song, go ahead/backwards one song, forward/rewind a song, browse the playlist. Each of these tasks are traditionally placed on buttons. Buttons are perhaps the only input mediums we engineers would imagine an electronic gadget to have. This is because the implementation of the music player (or any other electronic device) has been based on binary switch logic. But that is not what a user would expect. So the designers of the ipod used a wheel to hide the digital implementation to serve the analog semantics. Browsing should be as smooth a process as moving your finger is. The song skipping is however a digital command and a switch should suffice for that, so clicking the same wheel will serve the purpose.
Easiness is another aspect of ipod's UI. An interface should be intuitive. A gadget should have advanced features, but the user interface should serve only the most intuitive purpose first. So for advanced users few more controls could be stuck on the device. But that would be so ugly. Imagine having Up/Down buttons, Settings button, a button to turn on the backlight of ipod. Apple's solution to this problem is ingenious - Instead of providing these additional controls in space, hide them in time! You use the same buttons that you use for song playing, but for different length of time and voila! you uncovered an advanced feature of your ipod. It sounds so obviously 'intuitive'! Then why hasn't every gadget done this. That is because, bridging this gap between syntax and semantics requires fine engineering of the product. Time multiplexing several functions on a single hardware switch, needs carefully written code. The polling techniques to read the switch, the contraints on the size of the code - all need ace engineering. Only then, the meaning that the user expects gets conveyed to the gadget.
I've more thoughts on this issue, but 'll save them for some time in future. Till then, any comments are welcome.
We expect our gadgets, our software programs to do what we mean and not what we tell them. But this is a challenging problem. The means by which we can "tell" our gadgets what we want, are not sufficient to convey them what we "mean". This is the gap between syntax and semantics.
The innovations in User Interface strive to bridge this gap. Ipod is an excellent example of such innovation. The advances in the desktop GUI is another.
What is it that makes it so hard a problem? Or why is this a problem at all? Here are my 2 cents on this topic, in context of my Ipod.
Why is ipod's interface more user friendly than previous music players?
Let us first see, what we want a music player to do? - play a song, go ahead/backwards one song, forward/rewind a song, browse the playlist. Each of these tasks are traditionally placed on buttons. Buttons are perhaps the only input mediums we engineers would imagine an electronic gadget to have. This is because the implementation of the music player (or any other electronic device) has been based on binary switch logic. But that is not what a user would expect. So the designers of the ipod used a wheel to hide the digital implementation to serve the analog semantics. Browsing should be as smooth a process as moving your finger is. The song skipping is however a digital command and a switch should suffice for that, so clicking the same wheel will serve the purpose.
Easiness is another aspect of ipod's UI. An interface should be intuitive. A gadget should have advanced features, but the user interface should serve only the most intuitive purpose first. So for advanced users few more controls could be stuck on the device. But that would be so ugly. Imagine having Up/Down buttons, Settings button, a button to turn on the backlight of ipod. Apple's solution to this problem is ingenious - Instead of providing these additional controls in space, hide them in time! You use the same buttons that you use for song playing, but for different length of time and voila! you uncovered an advanced feature of your ipod. It sounds so obviously 'intuitive'! Then why hasn't every gadget done this. That is because, bridging this gap between syntax and semantics requires fine engineering of the product. Time multiplexing several functions on a single hardware switch, needs carefully written code. The polling techniques to read the switch, the contraints on the size of the code - all need ace engineering. Only then, the meaning that the user expects gets conveyed to the gadget.
I've more thoughts on this issue, but 'll save them for some time in future. Till then, any comments are welcome.
Subscribe to:
Posts (Atom)