[TUHS] Dennis' Draft of the Unix Timesharing System: not so draft?

[Dan Cross]

On Mon, Dec 19, 2016 at 3:59 PM, Clem Cole <clemc at ccc.com> wrote:
>
> On Mon, Dec 19, 2016 at 3:10 PM, Noel Chiappa <jnc at mercury.lcs.mit.edu>
> wrote:
>
>>
>> Not really a response to your question, but I'd looked at that
>> ​ ​
>> 'UnixEditionZero' and was very taken with this line, early on:
>>
>>   "the most important features of UNIX are its simplicity [and] elegance"
>>
>> and had been meaning for some time to send in a rant.
>>
>> The variants of Unix done later by others sure fixed that, didn't they?
>> :-(
>>
> ​One of my favorite comparisons and definitions of "bloat" came when I
> discovered years ago that the SVR3 >>boot<< system was larger than the V6
> kernel.
>

To be fair, I think some of the complexity is because hardware is more
complex now. It never ceases to amaze me how baroque some of Intel's stuff
has become.

On a related note, great as my respect is for Ken and Doug for their work on
>> ​ ​
>> early Unix (surely the system with the greatest bang/buck ratio ever),
>
> ​+1​
>
>
>
>
>> I have
>> ​ ​
>> to disagree with them about Multics. In particular, if one is going to
>> have a
>> ​ ​
>> system as complex as modern Unices have become, one might as well get the
>> ​ ​
>> power of Multics for it. Alas, we have the worst of both worlds - the
>> size,
>> ​ ​
>> _without_ the power.
>>
> ​Mumble -- Other than one important idea (single-level-store as you
> said), I'm not so sure.​  I think we ended up with most of what was
> envisioned, and some of the SW things (like the "continuation" model and
> how dyn-linking ended up working in practice) - I think we are ahead of
> Multics.   Winders more than UNIX (IMO) ended up with the complexity and
> bloat and most of the bad ideas without the good.  But I think UNIX mostly
> was able to stick to what was important (except for the loss of "small is
> beautiful" - my rant).  Some of the HW idea moved on - Intel picked up
> segments and rings. Look at INTEL*64, we use 2 rings and stopped using
> using segments because it too hard to program around them ---  both
> proved to be unusable/impractical when they were released.
>

Yeah. The only remaining vestige of x86 segmentation seems to be FS and GS,
which are often used for thread local storage.

(Of course, Multics made some mistakes - primarily in thinking that the
>> future
>> ​ ​
>> of computing lay in large, powerful central machines, but other aspects of
>> the system - such as the single-level store - clearly were the right
>> ​ ​
>> direction.
>
> ​I agree, and this may yet come back.   It's too bad too many of the
> younger engineers have not studied it.  I was recently reviewing some stuff
> from a couple of our younger Linux jockeys and they have re-invented
> something like it.   I smiled and said -- yes it >>is<< a great idea, but
> it has been done.​
>
>
>
>
>
>> And wouldn't it be nice to have AIM boxes to run our browsers and
>> ​ ​
>> mail-readers in - so much for malware!)
>>
> ​Indeed.​
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20161219/e51dac2e/attachment.html>