------------- Original Message --------------
Aseem <aseem(a)ports.cmc.net.in> wrote:
On Fri, 29 Jun 2001, N. Shrikant wrote:
> Are *.gz files really compressed into smaller sizes?
try yourself. on linux note the file size before
compression. and then then compress it with gzip, bzip2,
winzip. and you will find that all these utilities do
compress the files.
compression you get depends on file format.
i found that gzip compresses more for text files while
bzip2 was good for binary file compression. i found both
gzip and bzip2 better than winzip when it comes to maximum
amount of compression.
---------------------- End ---------------
Yep, you are right, compression results do depend
on the file format. Compression basically reduces
redundancies in data. Some file formats are
already in compressed form. Like JPEG, MP3 etc.
An MP3 file is just a compressed WAV file. So
compressing these file formats won't give you
very high compression ratio, since most of the
redundancies have already been eliminated.
Try saving a (fairly large) BMP as a JPEG
file. Notice the reduction in size, which is
a result of compression.
If, interested in compression techniques, you'll
find loads of reading material on the Net. If you
know someone from the engineering fraternity, try
getting hold of books like 'Computer Networks' by
Tanenbaum, or 'Principles of Comm. Systems' by
Taub & Schilling. Look for Information Theory in
the latter. These are the books I came accross.
There are many more.
Buy Feng Shui Package for Rs. 151/- only, at http://shopping.rediff.com/shopping/fengshui_mailer.htm
This is to remind you of the following ACM event.
Once again, sorry for the delayed email.
- Durgesh, for ACM Mumbai
Friday, June 29, 2001 - 6:30 p.m.
Title: Data Compression
Speaker: Sayyed S. A. A., NCST
Venue: Lecture Theatre, NCST, Juhu
Abstract:Data compression techniques are applied to files to reduce their
size, which in turn increases the capacity of data storage and speeds up
data transmission. Data compression draws heavily from information theory,
which deals with the probability of some symbol appearing in the
input. For totally random sources, each symbol is equally likely, but
random sources are uncompressible. This means that the more probable an
event, the less information it contains. Data compression techniques also
work on this principle. If we look at real-world files, we find a lot of
redundant data in them. Data compression exploits this redundancy to its
advantage. I intend to give an overview of data compression
About the Speaker: Mr Sayyed is a consultant in the Database and Knowledge
Engineering division at NCST Juhu. His interests include Databases,
compression and encryption.
Durgesh D Rao (DDR) | Email: durgesh(a)ncst.ernet.in
Research Scientist, KBCS Division | Ph: 6201606x372(w), 7563437(r)
National Centre for Software Technology | Fax: +91-22-6210139
Gulmohar Rd 9, Juhu, Mumbai 400049, INDIA.| Web: www.ncst.ernet.in/~durgesh
Hi this is a newbie here
I have 2 NICs both recognized by kudzu as RTL 8139 and RTLtoo. However
after i changed the ip address on eth0 it started to behave strangely.
Both the cards didnt ping their own ips neither the ips of the network
machines to which they were connected. Is there a prob with the cards or
with Linux generic divers?????
Pls help me out in this
Thanx in advance
> The main thing is that IISc will not be building the simputer. Only
> specification. The specification is released openly under the SGPL,
> anyone can build an implementation. The software is downloadable
> the GPL.
But, what I am trying to point out (unsuccessfully) is that, I admire
the things done at IISc (had interacted with one of the guys) but
based on the things done at IISc (including the true goal of reaching
the masses) companies like Encore are taking a free ride on getting
the government to subsidize certain operations of theirs, that is
really sick, beacuse for them its primarily business.
Any ways, just for a bit of info about Encore, they have a whole lot
of old 486 laptops lying with them which they are ready to sell off at
dirt cheap rates (remember my earlier post regarding the same?), that
just goes to prove the condition of the company...
Well, its their job, let them do it as per what they think is right.
> Anyway, I thought the project was good, and you didn't. Lets leave
> at that.
sorry if I gave an impression that I want to prove you wrong or that I
am right, just that I am really feeling sick about the whole issue of
> The product is an Open Source hardware project, which has naturally
> Open Source approach to its software development as well. The IISc
> undertaken the devlopment, with the active cooperation of many
> that it has reached the kickoff stage, it is possible to reach out
> commercialise the production. So Encore (as I understand from the
> do not recall seeing any announcment to that effect, but then I
> surfed thru the site either for a couple of fortnights) is one of the
> companies that has decided to do this.
Encore has not one of the companies that has decided to do this after
the project is over, rather Encore has had vested interests since
begining and has been investing on the hardware heavily.
So simply put, the situation is like:
Encore developed the hardware and the IISc guys developed or rather
put together the software, there are just a few instances where the
IISc guys have litterally written the device drivers (I had
interacted with one of them, a guy called Pai, don't remember is first
> If the ex-factory cost is 3k and the retail price is 6k, does anyone
> this is wrong?
Vickram, the ex-factory costs can't be 3K, just the base component
costs are 3K.
As in my earlier mail, add the rest of the stuff and voila, you have a
I have been browsing thru the debate on Simputer, and am curious about some
of the directions this has taken.
The product is an Open Source hardware project, which has naturally used the
Open Source approach to its software development as well. The IISc has
undertaken the devlopment, with the active cooperation of many others. Now
that it has reached the kickoff stage, it is possible to reach out and
commercialise the production. So Encore (as I understand from the mails, I
do not recall seeing any announcment to that effect, but then I haven't
surfed thru the site either for a couple of fortnights) is one of the
companies that has decided to do this.
If the ex-factory cost is 3k and the retail price is 6k, does anyone think
this is wrong? I very much doubt that the company will make any money at all
on this for many years to come. Unless of course, it takes off like a rocket
(a la i-mode in Japan) in which case I am sure that others will jump on the
bandwagon and the price will fall. Till then the 3k gross margin per piece
will flow out like water, absorbed by thirsty distribution, marketing and
Why am I so interested? As some of you will know from my earlier mail, I
have initiated the Hawking Communicator project in Open Source
(www.radiophony.com/hawking.html) and at some point this project too will
move towards commercialisation. I am keen to discuss all aspects of this
with as many groups of people as possible, so that we as developers are
properly prepared to do whatever it takes to get the software used. This is
not the forum for such a discussion, so please do join in the development
discussion at Yahoo!groups (/radiophony) or mail me directly.
>>Now tell me Philip, if you have used a handheld computer, is it
>>easier/faster to type or scribble?
>>Mind you, scribbling is not the same as writing.
>>And the hand writing recoqnition technology can never be perfected,
> Did you read about the stuff called "tapatap" that they describe on
> website? From what I gather, they have devised a very rudimentary
> mechanism for very simple interaction when requird. Keep in mind
> device is intended for _illiterate_ villagers, and nobody expects
> enter any data in it. That is mentioned explicitly in the wired.com
Now now Sameer, there you go again...
Have you ever worked on a handheld?
I too probably wouldn't have except for a rich friend of mine.
What do you think would it mean by a rudimentary input system?
Do you think just by placing a few taps you can access any data?
For accessing data, you need to place queries, and queries need
explicit search parameters, how do you suppose that can be achieved?
>>the best of technology companies like Apple, Palm and a few Russian
>>(who pioneered it) companies have tried and failed miserably.
> So WHAT? Even the GNU people were unable to create a kernel for
> system until a University student from Finland came along ...
Again, Sameer, have you ever thought about how handwritting
Have you thought or tried to code a handwriting recognition interface?
Do you know how difficult it is?
Or rather do you know a simple character like "A" can be written in
more styles than one by multiple people with different personality
That "A" is just from the Roman script, have you thought the
complexity of deciphering any of the Indian scripts (I would consider
Devnagari to be the simplest).
>>Have you figured out the BOM of the Simputer?
> I have to confess absolute ignorance in this regard!
>>Once you are through with the above, tell me if its possible to
>>a "Simputer" (as the creators have envisaged)
> People from the original Simputer team at IISc have established a
> commercial venture just to do that - called Picopeta something ...
>>One of the best projects in India??
> Reminds me of the fact that India was the first country in the world
> satellite communication for mass rural education, that too when
> didn't even have a single communcitation satellite of her own ...
> I think that what the IISc people are trying to do is to turn the
> of a "computer" on its head. They are trying to use technology to
> people in a way comfortable to them - this is a place where we have
> back all our presumptions about human-computer interaction based on
> Think of it as an interactive TV set if you want, designed to
> informatin rather than handle complex computing tasks.
Now what do _you_ think is the concept of a "computer"?
As you mentioned that "They are trying to use technology to help
people in a way comfortable to them", that is exactly what _Human
Computer Interaction_ is all about...
So we cannot keep back all our presumptions about Human Computer
> I had always hoped to get close to the Simputer project if I had
> to get into the IISc for my post-graduation; but I guess that won't
> happening now!
If you hoped to get to IISc, well thats good, thats a neat place with
lot of facilities...
But if you hoped to get close to the Simputer project, forget it, its
not worth it...
Just a bunch of techies don't make great computing devices, for that
you need to have experts on HCI and somebody with a clear vision about
what is to be done.
I have a DSL connection to the World. A hub is plugged into it, and
2-6 computers into the hub. Naturally, the computers have a class C
(?) between them, which should not be routable from the World. The
computers have their own connection to the World through the hub and
DSL. I think the whole thing is a star topology, at least for the
Is what I have described above correct, and possible? I suppose
ideally, one of the Linux boxen would get the DSL line and act as
firewall to the rest through the hub?
Now... whichever way, how do I do it on Linux? ipchains alone is
US-bound grad students! For pre-apps, see <URL:http://quickapps.cjb.net/>
Within five minutes, ReRob learned that walking through the basement of
Fuller Labs wearing futuristic body armor and pushing a variable-geometry
motorcycle raises eyebrows. Telling students that this is the true power
of UNIX didn't help. -- Undocumented Features I.