[TUHS] Lions book

[TUHS] Lions book

[Doug McIlroy]

I learned from Marianne Lions that she's still receiving
royalties from John's book. What a testimonial to both
John and the system!

Doug

Re: [TUHS] Lions book

[Clem Cole]

[-- Attachment #1: Type: text/plain, Size: 836 bytes --]

Indeed!!


FWIW I continue recommend it young hackers to read/examine/study before
they try to open something like the current Linux kernel.  The MIT xv6 work
 to put it on an Intel Architecture just adds to the value.

As the recent thread/argument on the merits/demerits of the use/abuse of
the preprocessor shows - simple and straightforward is easier to understand
regardless of the arguments wrt how helpful the macros can be.

6th Edition is clear and if you want to understand what it takes and how it
works, John's commentary it difficult to be.

On Fri, Jan 17, 2020 at 9:32 AM Doug McIlroy <doug@cs.dartmouth.edu> wrote:

> I learned from Marianne Lions that she's still receiving
> royalties from John's book. What a testimonial to both
> John and the system!
>
> Doug
>
-- 
Sent from a handheld expect more typos than usual

[-- Attachment #2: Type: text/html, Size: 1399 bytes --]

Re: [TUHS] Lions book

[Larry McVoy]

On Fri, Jan 17, 2020 at 09:50:24AM -0500, Clem Cole wrote:
> 6th Edition is clear and if you want to understand what it takes and how it
> works, John's commentary it difficult to beat.

It's a good starting point but it's pretty outdated.  I like to go on and
on about how much I love the SunOS 4.x kernel but it is outdated as well.

I wish there was a v6/SunOS like kernel that was as clean but had good
support for SMP and NUMA and TCP offload (and probably a long list of
other useful stuff I've forgotten).

Teaching kids how a single threaded kernel works is cool but it's
also misleading, the world has gotten a lot more complex.  And while
the kernels of decades ago were clean and simple, I don't know of
a kernel to point people to that has the clean code that SunOS had.
Solaris isn't it, though it has some bright spots.  Linux is meh, it's
better than nothing by a lot but I would not point to it as "read this,
kid, you'll see the architecture".  It's not clear there is a good
answer.

Re: [TUHS] Lions book

[John P. Linderman]

[-- Attachment #1: Type: text/plain, Size: 1121 bytes --]

I heard from her as well. So far, no immediate fire danger where she lives
in Australia, but her garden (which may be Australian for lawn) is dying
because of water restrictions.

On Fri, Jan 17, 2020 at 9:52 AM Clem Cole <clemc@ccc.com> wrote:

> Indeed!!
>
>
> FWIW I continue recommend it young hackers to read/examine/study before
> they try to open something like the current Linux kernel.  The MIT xv6 work
>  to put it on an Intel Architecture just adds to the value.
>
> As the recent thread/argument on the merits/demerits of the use/abuse of
> the preprocessor shows - simple and straightforward is easier to understand
> regardless of the arguments wrt how helpful the macros can be.
>
> 6th Edition is clear and if you want to understand what it takes and how
> it works, John's commentary it difficult to be.
>
> On Fri, Jan 17, 2020 at 9:32 AM Doug McIlroy <doug@cs.dartmouth.edu>
> wrote:
>
>> I learned from Marianne Lions that she's still receiving
>> royalties from John's book. What a testimonial to both
>> John and the system!
>>
>> Doug
>>
> --
> Sent from a handheld expect more typos than usual
>

[-- Attachment #2: Type: text/html, Size: 1955 bytes --]

Re: [TUHS] Lions book

[Warner Losh]

[-- Attachment #1: Type: text/plain, Size: 1653 bytes --]

On Fri, Jan 17, 2020 at 8:24 AM Larry McVoy <lm@mcvoy.com> wrote:

> On Fri, Jan 17, 2020 at 09:50:24AM -0500, Clem Cole wrote:
> > 6th Edition is clear and if you want to understand what it takes and how
> it
> > works, John's commentary it difficult to beat.
>
> It's a good starting point but it's pretty outdated.  I like to go on and
> on about how much I love the SunOS 4.x kernel but it is outdated as well.
>
> I wish there was a v6/SunOS like kernel that was as clean but had good
> support for SMP and NUMA and TCP offload (and probably a long list of
> other useful stuff I've forgotten).
>
> Teaching kids how a single threaded kernel works is cool but it's
> also misleading, the world has gotten a lot more complex.  And while
> the kernels of decades ago were clean and simple, I don't know of
> a kernel to point people to that has the clean code that SunOS had.
> Solaris isn't it, though it has some bright spots.  Linux is meh, it's
> better than nothing by a lot but I would not point to it as "read this,
> kid, you'll see the architecture".  It's not clear there is a good
> answer.
>

It's but the first step on the road to understanding. I'd been working on
the FreeBSD kernel for years when I re-read the Lions book. The stark
simplicity of the v6 kernel helped everything suddenly 'click' into place
in the code I was reading in the FreeBSD kernel, even with 30ish years of
changes to the v6 code base that lead to the FreeBSD kernel...

Newer systems are a lot more complicated. And they need to be to get the
full performance out of the system. Yet understanding the basics without
the extra clutter has great value.

Warner

[-- Attachment #2: Type: text/html, Size: 2156 bytes --]

Re: [TUHS] Lions book

[Clem Cole]

[-- Attachment #1: Type: text/plain, Size: 2875 bytes --]

I completely agree.  Studying and >>learning<< from V6 allows you to
consider the basics of what any good OS, no just a UNIX-like system, has to
do provide simple but complete services.  Examining V6 not most of the
things you find today removes a lot of the noise, like threading, parallel
execution, vm, networking, which are fine topics for later, but Warner is
right - get a solid understanding first.  It also helps to understand what
makes 'UNIX-ness' and why it was different from anything before it came on
the scene.

BTW: It' why I still like Pascal (Delphi) over C or C++ as a first language
(I admit, I'm leaning towards Go these days, but Go lacks a good teaching
text).  This is what I heard Doug saying.   IMO:  Lion's book and the V6,
can be considered 'old' by contemporary standards, but they are still 100%
appropriate and because the book and code is so simple, the teacher and the
student can focus on what really matters (*i.e.* learning to walk carefully
in a directed manner and get to your destination before you are forced to
run with the bulls and avoid getting run over).

On Fri, Jan 17, 2020 at 11:01 AM Warner Losh <imp@bsdimp.com> wrote:

>
>
> On Fri, Jan 17, 2020 at 8:24 AM Larry McVoy <lm@mcvoy.com> wrote:
>
>> On Fri, Jan 17, 2020 at 09:50:24AM -0500, Clem Cole wrote:
>> > 6th Edition is clear and if you want to understand what it takes and
>> how it
>> > works, John's commentary it difficult to beat.
>>
>> It's a good starting point but it's pretty outdated.  I like to go on and
>> on about how much I love the SunOS 4.x kernel but it is outdated as well.
>>
>> I wish there was a v6/SunOS like kernel that was as clean but had good
>> support for SMP and NUMA and TCP offload (and probably a long list of
>> other useful stuff I've forgotten).
>>
>> Teaching kids how a single threaded kernel works is cool but it's
>> also misleading, the world has gotten a lot more complex.  And while
>> the kernels of decades ago were clean and simple, I don't know of
>> a kernel to point people to that has the clean code that SunOS had.
>> Solaris isn't it, though it has some bright spots.  Linux is meh, it's
>> better than nothing by a lot but I would not point to it as "read this,
>> kid, you'll see the architecture".  It's not clear there is a good
>> answer.
>>
>
> It's but the first step on the road to understanding. I'd been working on
> the FreeBSD kernel for years when I re-read the Lions book. The stark
> simplicity of the v6 kernel helped everything suddenly 'click' into place
> in the code I was reading in the FreeBSD kernel, even with 30ish years of
> changes to the v6 code base that lead to the FreeBSD kernel...
>
> Newer systems are a lot more complicated. And they need to be to get the
> full performance out of the system. Yet understanding the basics without
> the extra clutter has great value.
>
> Warner
>

[-- Attachment #2: Type: text/html, Size: 3929 bytes --]

Re: [TUHS] Lions book

[Adam Thornton]

> On Jan 17, 2020, at 8:23 AM, Larry McVoy <lm@mcvoy.com> wrote:
>  I don't know of
> a kernel to point people to that has the clean code that SunOS had.

It’s not exactly a kernel and it’s not exacly … well, anyway:

selfie is pretty cool.

http://selfie.cs.uni-salzburg.at/

Adam

Re: [TUHS] Lions book

[G. Branden Robinson]

[-- Attachment #1: Type: text/plain, Size: 295 bytes --]

At 2020-01-17T10:21:58-0700, Adam Thornton wrote:
> It’s not exactly a kernel and it’s not exacly … well, anyway:
> 
> selfie is pretty cool.
> 
> http://selfie.cs.uni-salzburg.at/

Oh, heck yeah, it is.  Every bullet point makes my heart go pitter-pat.

:D

Regards,
Branden

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]