[unios] Re: Networked GUI (was Posix and Networks)

[Srikant Sharma (Chiku)]

From: "Srikant Sharma (Chiku)" <srikants@wipinfo.soft.net>

On Thu, 17 Dec 1998, Pieter Dumon wrote:

> 
> > >> Oh yes. I see no need to use the same program (as long as the effect is
> > >> equal, the method dones't matter), but sharing data is very useful. Just
> > >> have all users open the same file in a special shared mode, and they are
> > >> notified when changes are made by others and what's changed (changes are
> > >> seen directly). This requires little effort from the application
> > programmer.

Isn't this standard mmap() interface provided by *ANY* unix filesystem ?

> > >
> > >I think it shared programs would be useful too.
> > 
> > I maybe think of this different from you. The binary program could always
> > be shared, but that's just so evident. What I mean is that as long as the
> > program can read the data and update when someone else makes changes, what
> > program you use doesn't matter.

This is the basic concept of memorymapping used in dynamic libraries.
In unix (solaris specifically) .. all the required  dynamic libraries are mapped
in the memory at the run time and the code is picked up runtime.

the mmap can work over nfs also ....

> 
>  
> > >> I have another suggestion. Redundance in networking, which allows one or
> > >> several nodes to fail without loss of the joint computation or information.
> > >> To how big degree this can be done by the OS I don't know, but it's
> > >> certainly useful. What if you could just declare some data to logically
> > >> exist on the network, and it's automatically mirrored on several nodes,
> > >> with accesses made to the nearest copy without any effort from the
> > >> application's side? 
> > >> binEng

This is really a BIG thing ...... what you are asking for is 100% fault 
tolerance .....  The servers from Tandem have it ....
reduced form of this is reliability .... (fail proof ... but slight 
performance degradation because of failures ...) NCR servers give this to you .

This requires an extensive hardware support ... this can't be simply done in
software alone ... And no need to state .... if the hardare is special ...
u need to design special software for it .. :)


> > >> 
> > >
> > >Hm. What you are referring to, is ont of the buzz-words of the last year,
> > > CLUSTERING. Unix can do it, MS just starts to get some in-the-field
> > >experience. It's really important for a corporate server-OS.
> > 
> > Yep. Should be done as something like a virtual device driver. Doesn't need
> > to be done from the very beginning.
> 
> No, but it's pretty hard to implement correctly and foremost crash-proof.
> MS developpers are really not _that_ bad. It just takes some time to
> implement it.

just another view (could be similar )

> To give an example of how long it takes to get an OS _really_ good:
>  Unix is now 30 years old. It started in 1969 as a single-user
> multitasking system, then it was made multi-user. Now it's the most
> capable OS in the world.

MVS .... a giant OS .... it's used when unix fails to give the 
required robustness ...  (disclaimer : I have never used MVS. But I 
have heard about it several times)

> Scalability is not something that was invented

very true ...
(this reminds me , Pat .. I have some comments on your defination of
scalability .. will send them over to you soon ...)

> as a buzz-word to obstruct MS. No, it is what Unix is and what NT is not.

Not every (flavor of )UNIX ......

> Unix can run on very small integrated systems, as well as PCs, as well as
> multiprocessor servers a well as CRAYS etc. It can run on all cpus you can 
> imagine. But still, it takes Unix lots of time to be ported stable.
> Unix isn't explicitly 32 or 64-bits, and while it is fairly easy to

It is ......  when you say an OS is 32 bit ..... The kernel *has* to be
32 bit ... when you say 64 bit the it has to be 64 bit ....

soalris 7  (or 2.7) is exactly 64 bit .... All drivers you write need to
be 64 bit drivers ..  No kernel code can be 32 bit in a 64 bit OS ..

It is a different matter that there are libraries which act as bridges 
for compatibility of older applications  ....


> implement a new Unix kernel for, say the new 64-bit Merced that's coming
> up, it is fairly hard to make it also _stable_ and optimized. MS is

Actually it is failry hard to make *anything* fairly stable  and optimized.
This is true of every field in the word. Not just OS.


> allready working on a 64-bits NT for the Merced. It will take them years
> to get it stable, but they will realease it before it is stable (Stable is
> not what I call NT, it's what I call Unix), allthough the NT design would
> make it easier to port NT to a new machine than Unix. It's just that MS
> doesn't have experience with something other than x86. Also, Unix
> corporations are spending much more time on their OS before it is released
> than MS. I can't believe it MS is doing so stupid. You must know that
> they once were the manufacturer of one of the most popular Unices, namely
> XENIX. Do you know some people are really starting to fear for MS? Not

well the MS caters to a different market segment .... The needs are different ..
And XENIX is not MS brainchild .... they owned it is totally different.


> based on speculations, but on facts! Some people predict MS won't last  
> for long anymore ...  I don't know. They got big money and can survive
> lots of things, but they are now being attacked on their own field:
>  - Linux is being made more user-friendly, whike retaining its power 
>  - Samba on Linux can now run NT domain servers etc, so you can build a
>    complete NT network without using NT, and integrate Win95 PC's or so.

This is actually a matter of implementation of protocols ...

>  - Linux also runs Netware, so Linux runs NDS! Bye,bye Active Directory.
>   
> > >An OS that can run _really_ non-stop (off course on the best hardware),
> > >should be able to be implemented too. Although... Tandy and some others
> > >seem to be the only firms able to do this... So it is rather difficult, I
> > >think.   
> > 
> > No reboots? That is a good goal. The final test of stability.
> 
> It's off course only achievable on Tandy-like machines. But with no
> reboots I mean NO reboots. All lesser Unix systems have to reboot some
> time in their lifetime. Tndy Unix systems don't. (QNX is just like that)


>  
> Pieter
> 
> ----------------------------------------
>  Pieter.Dumon@rug.ac.be               
>                                       
>  http://studwww.rug.ac.be/~pdumon     
>  
>  ICQ  : 12428974
> ---------------------------------------
> 
> ------------------------------------------------------------------------
> To unsubscribe from this mailing list, or to change your subscription
> to digest, go to the ONElist web site, at http://www.onelist.com and
> select the User Center link from the menu bar on the left.
> ------------------------------------------------------------------------
> UniOS Group
> http://members.xoom.com/unios
> 


------------------------------------------------------------------------
To unsubscribe from this mailing list, or to change your subscription
to digest, go to the ONElist web site, at http://www.onelist.com and
select the User Center link from the menu bar on the left.
------------------------------------------------------------------------
UniOS Group
http://members.xoom.com/unios