Hello Group, Hello Sudhanshu, Hello Kapil Jain,

Thanks for your mails.

Hope my mail doesnt seem like deviation from topic- but am taking liberty to
share my experience.

I was double-minded about writing long mails-- but given my own interest-- I
am writing in--- and if there are a few people (say atleast 2 or 3)-- I am
happy to give some of my time-- and we can even have a kernel
hack/discussion session sometime.

Basically, you need to do things step-by-step to get exposure-- because many
things are different from impression you'll get from books-- and until you
yourself spend some weeks digging in-- till then, you will risk getting
stuck and never finishing a project.

>
>Please Advise, though the above posts were very useful ,thanx guys.
> Has someone tried to understand these project oriented courses on OS
> from renouned Universities like OS/161 , pint OS and xv6. ? At present
> my examinations are going on, so could not devote  time in R&D.   If
> someone tried anything ,please share ur experience and your
> knowledge.
>
> This project of OS development requires people , who can contribute
> there skills. If this was done in Berkeley then why cant it be done
>here again with latest technology and Web support?
>
Have you had a chance yet to compile the OS and also re-build the toolchain
(gcc, bin-utils etc) from sources ?

If not, I suggest you do that. Dont just re-build the kernel; also rebuild
bin-utils; gcc and the entire tool-chain from sources-- the first time I did
that (years back)-- the results and problems were beyond what I expected--
and doing this will teach you a lot.

Also, there is a book by Tannenbaum, about OS Design and Implementation--
which has entire source code of minix os-- and which I suggest you work
through-- because many end of chapter exercises will expose you to many
interesting concepts-- which otherwise it might take you months or years to
get exposed to.

The book I am mentioning has the entire source code of minix operating
system-- which was even Linus Torvalds' starting point.

You should try to build that source code
---

A lot of things look doable-- and they are-- but I suggest you do things
step by step, deepening your own knowledge and exposure before you go
deeper-- because many times one ends up re-inventing the wheel and doing
things that others have already done.

You should gain exposure and experience with different things-- so that when
you embark upon your own journey of making a OS-- you produce results.

I have also sent you a unicast mail-- respond to that when you get a chance.

I am happy to help and/or share knowledge-- so feel free to write to me--
either on list OR unicast.

>
>Important: I think microkernel approach is quite reliable and feasible
>at initial stages, of OS development. But Linux kernel is monolithic
>and its kernel code is quite large. So I suppose , one should go with
>both microkernel and monolithic kernel for combined
>functionalities..i.e a hybrid kernel would be a best choice.
>
even this topic can be food for lots of discussion-- and time and
opportunity permitting-- it will be good to someday dig deeper and share
thoughts on this.

Right now, I am tempted to make remarks like "because xxx therefore yyy is
not good even thought it seemingly offers advantages aaa, bbb, ccc"--

but yes-- such statements make sense only if one has the time and energy to
qualify and quantify one's generalizations-- otherwise it's just random
argument-- hence suggesting you deepen your own knowledge and exposure.

Regards,

Nalin


On Sun, May 23, 2010 at 6:57 PM, Sudhanshu Shekhar Sharma <
[email protected]> wrote:

<snip>

-- 
l...@iitd - http://tinyurl.com/ycueutm

Reply via email to