Talk:Distributed Systems

This textbook is based on class notes from a graduate-level comparative operating systems interfaces course.

Guidelines and Direction
Each chapter should follow this pattern:

== chapter == === section == ==== exercises for section ====

Answers for the exercises belong in the (currently empty) answers appendix.

Generally, I suggest we keep the number of chapters small, but have lots of sections.

There's a C programming appendix at the end that covers some stuff people might not know about C.

pseudocode vs. real code
While functional examples are great, sometimes it's best to introduce a topic with psuedo-code.

Additionally, you may be interested in Beej's Guides, which I've found to be particuarly useful.

--Eibwen 06:59, 1 Oct 2004 (UTC)

Thanks for the Beej link! Another site I really like is this guy's, and he covers just about everything.

--Waxmop 02:36, 2 Oct 2004 (UTC)

As an experienced system programmer, functional code is best. "Pseudo-code" is basically meaningless when the objective is a simple example of an API IMO.

Named pipes and locks
Consider adding sections for named pipes and locks (eg sephamores, mutex). Granted the latter you'll probably cover in the threads section, but locks should at least be their own subsection.

--Eibwen 17:56, 25 Sep 2004 (UTC)

Yeah, I just added some example mutex code today. I'm typing this stuff in as the class progresses.

goals for this book
I would consider naming this something else than "Distributed Systems". That is far too general term for Unix thread/process intercommunication software development and TCP/IP protocols. If you really want to make a book on this subject, its not just Ethernet/TCP/IP-based distribution (message passing) - what about CANopen, Lin, Profinet, EPL, etc. Also, we are not talking about sensor networks, control systems, ...

--Mikko Laakso

I agree that *currently*, this book seems to be almost entirely about multitasking on a single box, and "distributed systems" is a much larger field.

However, I really do want a book on distributed systems. So the stuff you see on "multitasking on a single box" is merely one chapter of what I hope this book becomes.

How can I make clear that this is still a rough draft, and there are still a lot of other sections yet to be written? How can I persuade people to write those other sections?

p.s.: There currently seems to be a lot of overlap with Computer programming/Concurrent programming Parallel Computing and Computer Clusters. So I expect that someday, after both have matured a bit, that several pages from this book will be moved into those books, and a few pages from those books will be moved into this "Distributed Systems" book. --DavidCary 05:44, 20 October 2007 (UTC)

I still think this book needs a better focus to be really useful. That would help and steer writing this book as well. Focus on one, more limited, subject instead scraping the top of everything. And perhaps later, would be good to make it clear for everybody by re-inventing the name.

--Mikko Laakso

As an experienced system programmer I think this book is great. If you want to rename go for it but the content is perfect.

-- A PHD EE User in Florida, USA

high and low memory
may be unclear --Remi (talk) 02:58, 25 February 2008 (UTC)