[comp.sys.att] SV R3.2.2 vs. R4.0

rembo@unisoft.UUCP (Tony Rems) (10/13/90)

In article <2152@megadon.UUCP> holsberg@pilot.njin.net (Peter J. Holsberg) writes:
>In the docs for SV/386 R3.2.2, I have read about the inclusion of
>stuff from Xenix and (I believe) BSD.  In the ads for R4.0, I have
>read of the same things.  Is this true?  Does it mean that R4.0 is not
>much of an advance over 3.2.2?  
>
>Thanks,
>Pete
>-- 
>Pete Holsberg	Mercer County College  Trenton, NJ
>The College on the Other Side of Route 1

No, that's not what it means.  It just means that, while AT&T has
made it's kernel large and unwieldy, it has attempted to include
all the features that both BSD and System V users are used to. 
Notable features are the vfs (virtual file system) which allows
you to use S5fs, ufs, or bfs yet write your code independently of
what file system you're using.  It includes system calls like 
statvfs which return a *generic* superblock so that you don't
have to dig around on the disk yourself.  Also, you have all
the BSD and S5 IPC mechanisms, and all the S5 and BSD commands.
So, no matter what you're used to, most of the features should
be there.  Now, before you start calling AT&T for your copy,
remember that, to include all this, as you can imagine, you 
end up with a huge kernel and a huge system.  Also, AT&T didn't
include support for multiprocessing.  So, the deal is, you can
get most of the things you want, but you'll have to pay for them.

Oh, also, AT&T, in it's effort to contribute to ease of use has
decided not to include on-line man pages in this release either.

Ah, the joys of Ma Bell...

-Tony

hjespersen@lion.uwaterloo.ca (Hans Jespersen) (10/20/90)

In article <3162@unisoft.UUCP> rembo@unisoft.UUCP (Tony Rems) writes:
 
>No, that's not what it means.  It just means that, while AT&T has
>made it's kernel large and unwieldy, it has attempted to include
>all the features that both BSD and System V users are used to. 

I have heard many people call the SVR4 kernel large but I am
not convinced that it is. Of course if you load all the options
particularly the Berkley Compatibility Package and all the 
networking stuff you are going to have a larger kernel than if
you don't, but it is still close to a comparably equiped Rel. 3
kernel. What I'm trying to say is if you are really concerned
about having  a small kernel, don't load all the packages or
if kernel size is such a big issue for you go back to running
Version 7 or some nice compact proprietary (yuk!) operating
system.

>Notable features are the vfs (virtual file system) which allows
>you to use S5fs, ufs, or bfs yet write your code independently of
>what file system you're using.  It includes system calls like 
>statvfs which return a *generic* superblock so that you don't
>have to dig around on the disk yourself.  Also, you have all
>the BSD and S5 IPC mechanisms, and all the S5 and BSD commands.
>So, no matter what you're used to, most of the features should
>be there. 

Exactly why SRV4 is such a good product for an inductry that
has been asking for a unified Unix operating system for years.
(even if it was as "large and unwiedly" as some people think
it is)

>	 Now, before you start calling AT&T for your copy,
>remember that, to include all this, as you can imagine, you 
>end up with a huge kernel and a huge system.  

But you don't have to include all this if you don't want to.

>                                             Also, AT&T didn't
>include support for multiprocessing.  

Be careful here. AT&T USL (Unix System Labs) has not released
a standard multiprocessing release yet ('91 I think) but
AT&T Data Systems Group has been selling mutiprocessing 3B2's 
for years and has just released a Symetric Multiprocessing '486 system 
running (you guessed it) System V Release 4 (4.0.3 actually).


>                                      So, the deal is, you can
>get most of the things you want, but you'll have to pay for them.

This is a very puzzling statement considering the commercial
nature of the computer industry.
 
>Oh, also, AT&T, in it's effort to contribute to ease of use has
>decided not to include on-line man pages in this release either.

I was under the impression that AT&T will make man pages available
(for SVR4) as an option to those who want it. Anyone know for
sure about this issue?
 
>Ah, the joys of Ma Bell...
>
>-Tony

-- 
Hans Jespersen