Core Mac OS X and Unix Programming

Michael McCracken writes "Finally, a new OS X programming book that isn't just another introduction to Cocoa. This book adds a lot to the available references by covering the system as a flavor of Unix, presenting information on important topics like sockets, multithreading and pipes, which other OS X books leave out. It also includes coverage and sample code for some of the unfamiliar new technologies that have been introduced recently, such as the Keychain, Rendezvous (aka Zeroconf), and using the Security framework to authorize users." Read on for the rest of his review.

Core Mac OS X and Unix Programming
author	Mark Dalrymple and Aaron Hillegass
pages	541
publisher	Big Nerd Ranch
rating	9
reviewer	Michael McCracken
ISBN	0974078506
summary	A developer's guide to the Unix foundations of Mac OS X, including coverage of recently added technologies. Includes complete source code and online companion material.

If you've been learning Mac OS X Cocoa programming, you might already know Aaron Hillegass through his excellent book Cocoa Programming for Mac OS X, which was one of the first good introductory books on the topic, and is still one of the best available. Information about this earlier book can be found at bignerdranch.com/Book/. Both Aaron and Mark are instructors at the Big Nerd Ranch, which offers courses in Mac OS X programming. More information about them and the courses can be found at http://www.bignerdranch.com/Company/Who.html. This book is based on the course with the same name at the Big Nerd Ranch. The book's website and a link to order it online can be found at borkware.com/corebook/ . Discussion and further information for both books can be found at cocoadev.com/index.pl?CocoaBooks.

Audience and Writing Style

This book is not an introduction to programming on OS X. It doesn't explicitly cover how to use Apple's Project Builder or Interface Builder, or much of the Cocoa or Carbon APIs, except during discussion of code examples. So if you're entirely new to programming or to using Mac OS X, start with a different book such as Hillegass' earlier Cocoa Programming for Mac OS X to get up to speed with using the development environment. This book will leave you behind at times if you are unfamiliar with using the command line, however, the examples are complete enough to follow along by just typing in what's in the book.

Core Mac OS X and Unix Programming does have some very basic material in its first few chapters. They focus on the details of C programming, using the compiler, memory management, and debugging. These chapters will be mostly review for anyone who's developed in C on Unix before, but will be invaluable for programmers who learned to program using Java, for instance. They should also be required reading for programmers who started programming with Objective-C and Cocoa and are still unsure about using "plain" C. If you've ever complained about having to use a C API from CoreFoundation in your nice pure Cocoa application, don't avoid this book -- you need to read it even more.

The book is clearly written and easily understood. The writing is occasionally conversational, in keeping with its history as a course textbook. In the grand history of well-written technical books, it is also occasionally funny. The authors don't try to sell the technologies they discuss, instead giving practical advice that's useful to a programmer who is trying to actually build something. For example, the authors discuss bugs and inconsistencies in the system, clumsy API design and other problems that aren't great ad copy but you will need to know to develop robust applications.

I found this aspect of the book one of the most appealing, that it felt as though I was actually getting down to business. Gems of practical advice that can cut short frustrating problems appear throughout the book, so be sure to read carefully, don't just skim.

Hits

Here I'll discuss a few examples of where I think this book really shines. First, the level of detail of the standard Unix APIs and the development tools is excellent -- I learned many immediately useful things in the first 13 chapters. For example, chapter 8, "Debugging With GDB," was not simply a repeat of the online help, but also contained useful tips about how to use GDB more effectively, from using Objective-C specific features to tracking subtle memory errors. Programmers who had only used Project Builder's interface to GDB will benefit greatly from this chapter.

Next, there is pervasive sample code. Each chapter had a complete sample program demonstrating the topic at hand. Much of this code is also available online: see "Online Supplements" below.

Finally, as I mentioned before, the text contains tips and reminders throughout about potential mistakes, tricky problems, and differences between Mac OS X, OS 9, and other Unix flavors. A particularly useful example of this is in chapter 24, "CVS." There is a small but important paragraph that discusses using CVS 1.11 (as used in Sourceforge) with Interface Builder .nib files that can really save some grief. In other chapters, they even include workarounds for system bugs in some of the sample code. This pragmatic approach is really appealing.

Misses

In this section I'll mention a couple potential disappointments. You will have to be willing to learn by just reading code at times. Most of the code examples are not explained line-by-line as is the custom in tutorial books. Comments explain tricky code, and the text covers the areas most relevant to the chapter topic, but for other sections, understanding is up to you. I mentioned this because although I feel the code is clear and a fair trade-off was made to fit in a lot of information, the amount of explanation you like is a matter of personal preference, and so you should know what to expect.

Pointers to further reading is another problem. Aside from referring to man pages, there is little attempt to point to good external documentation on any of the more complicated topics. One chapter is not enough to completely cover BSD Sockets, for example, and so a reference to a Unix network programming book would be useful. In fact, every chapter could be improved by a references section, even if it only collected links to Apple online documentation or Unix community websites. With all the practical knowledge in this book, the lack of clues on where to look to answer your own questions was disappointing.

Finally, the cost of the book, at $97.95, is higher than you might expect. I admit that as a student, I would have to think twice about paying this price, although I am sure it would be easily justifiable for professional programmers. I believe that it is worth the price, however, because you would have to buy several other books to cover the same range of topics, and you still wouldn't get the Mac OS X specific information.

Online Supplements

The authors have set up a promising resource for the book at http://borkware.com/corebook/ . The site includes the sample code, errata, reader comments indexed by the chapter and topic they refer to, and a general discussion board. There are already some errors listed, and a few pointers to useful documentation and interesting external discussions on mailing lists. The sample code is not complete at the time of this review, but more is being added. This site looks like it will be a useful addition to the book, especially if many good chapter-indexed comments are added. The site could be kept open as a reading companion while going over a chapter in the book. This site's organization is, in my opinion, much more useful and usable than other books' companion websites, including the site for "Cocoa Progamming," which hid its information from you unless you knew which page number was relevant to your topic.

Conclusion

Core Mac OS X and Unix Programming is a very useful book, and even if you've been developing on Unix systems for years, you can probably learn a few immediately useful things by reading it. I recommend it for any serious Mac OS X programmer who wants to know what to read next after all the tutorials that have come out in the last year or so. I suspect it'll become a canonical reference, and may even be in need of a clever nickname. Congratulations to Mark and Aaron on a job well done.

---

Michael McCracken is a grad student and Mac OS X developer; he says "I have not attended any Big Nerd Ranch courses, nor have I met either author, although I did see Aaron Hillegass in a crowd once." Update: 07/02 17:36 GMT by T : According to publisher AtlasBooks, bn.com won't actually be carrying this book, but you can get it right now from Atlas. Slashdot welcomes readers' book reviews -- to see your own review here, read the book review guidelines, then visit the submission page.

Whoa

Be careful, you might actually teach them some good programming techniques.

Microsoft port

It is interesting to note that Microsoft in a previous article said that they had no intention of porting Office to Linux right now, but being that they have ported it to OS X most of the work has already been done.

Re:Microsoft port

MS Office is a Carbon app. There's no carbonlib for any *nix except Mac OS X out there. Porting it to Linux would mean either porting carbonlib (i.e. a cleaned up version of the legacy Mac OS api + extensions) to Linux, or rewriting it from the ground up (just like the Mac OS X version is almost a complete rewrite compared to the Windows version).

Re:Microsoft port

Worse than just a Carbon app. MS Office for Mac is written using WLM (Windows Compatibility Layer for Mac - pronounced "will am"). This was Microsoft's idea for single sourcing office for Mac and Windows back in the early 90s.

They have long since given up the single source idea. The Mac BU has their own copy of Office Source. However, it is still written using bastardized Windows API on top of a Carbonized WLM.

The whole thing is a mess. There is basically a zero chance that they are going to pick that up and port it to Linux. If they decided they wanted to port it to Linux, they would be better off buying something like ThinkFree Office to be the basis of the port (for several reasons). In fact, *if* they were going to port to Linux, they'd probably do just that.

I used to be a Mac developer there.

Re:Microsoft port

Microsoft in a previous article said that they had no intention of porting Office to Linux right now, but being that they have ported it to OS X most of the work has already been done.

Actually that's not quite true. Microsoft programmed Office for the Mac using the Carbon API and libraries which are part of MacOS. This is a set of APIs which are completely proprietary to MacOS and which bring a program no closer to Linux than a program which uses the Windows API program does.

The only way in which doing a MacOS port of a program brings it closer to doing a Linux port is most programmers doing a port tend to separate out the parts of the program which rely on a particular operating system from the parts that are platform-agnostic. Thus you will most likely end up with a large chunk of the code which can be re-used to add a front end for any particular system type. This is not a requirement for a port however, it is just smart programming. In fact, it is even smarter programming to do this in the first place. Microsoft may or may not have done this, but I'm doubting that they did. From what I understand they basically program the Mac versions of their programs from nearly the ground up.

In programming the Myth series of games, it was often said by the developers at Bungie that the platform-agnostic parts of the games took up about 90% of the code, while each port took up 10% more. So for a 10% investment in additional coding you could sell the game to a another platform. This requires a bit of planning but it is a much better way to program than to do one version for one platform and then have to completely redo you work to get it to run on another. Finally, there is another great advantage to doing multi-platform programming. Often a bug which doesn't show on one platform will show up on another, allowing you to clean up any possible problems before they get you into trouble later on.

keychain

keychain has been a part of the mac os since OS 8, albeit not it it's current unix-ized form

Re:keychain

albeit not it it's current unix-ized form

So how did the keychain suddenly become "unixized"? It's still a part of the Carbon API.

Does

OSStatus KCFindAppleSharePassword (
AFPServerSignature * serverSignature,
StringPtr serverAddress,
StringPtr serverName,
StringPtr volumeName,
StringPtr accountName,
UInt32 maxLength,
void * passwordData,
UInt32 * actualLength,
KCItemRef * item
);

look unixized to you?

C'mon!

You can't be "unfamiliar with the command line" and a programmer. Pick one.

Re:C'mon!

If anything, people who don't use a command line are smarter than you because they're cutting down on keystrokes, thus improving their productivity.

Oh, they're so cute when they're first learning!

Seriously, if your idea of "productive" is "uses fewer keystrokes", then your teacher needs to boot you in the butt. Grow, learn, expand your mind.

$97.95?!

That'd put a major dent in my crack usage...

Aaron Hillegass as an instructor

I am approaching the end of Aaron's previous book, Cocoa Programming for Mac OS X, and can say that throughout, it's as though he is sitting there beside you, casually instructing you as you move through the work. An excellent introduction to Cocoa, it has given me a fairly solid graps of the concepts that make up Cocoa development (which began in the late 80's with NeXT) and I have made some real strides on my own, veering from the courework in the book. Goes far beyond some of Apple's cryptic guides I've encountered.

Oh...and do yourself a FAVOR and download Cocoa Browser [nifty.com] before you even lay down a single line of Objective-C. The ONLY way to access the frameworks references.

blakespot

Just wondering...

Those WWDC people who have access to Apple's upcoming OS, Panther...

How relevant will this information be with Panther merging to BSD 5.0 userland and the new Xoode environment?

I can't seem to justify the price for this book ... yet.

Confused

I see Unix in the title but there is no mention of SCO. Just what in the hell is going on around here?

Good News...

... If you have a Barnes & Nobel reader's advantage card:

List Price: $97.95
Our Price: $78.36
You Save: $19.59 (20%)

Readers' Advantage Price: $74.44

(OUCH!) This looks to be one book I'm going to have to skip. Bummer.

Wrong URL to Buy the Book

You can't (currently) get the book at bn.com. Use this web page to buy the book: http://www.atlasbooks.com/marktplc/00981.htm [atlasbooks.com]

...SCO lawyers start your engines...

will they sue these guys for having Unix as part of the Book title?

the price

I plan on buying this book when I get the cash, but from talking to Aaron last week at WWDC, the price is that high is because it's basically the course material for the classes they have at the Big Nerd Ranch. Of course...I can't afford the classes, so I'm going to get the book.

price is $104 -- am I making a mistake?

Am I reading it correctly that the price is $US97 plus shipping?
Also, is it showing up on Amazon? I looked and didn't see it.
Clues gratefully accepted...

Villagers need jo2y.

The book lacks sufficient references to 'jo2y',
I could only find about 3 or 4.

Weird Times

I know OS X is all unix central nowadays...Yet to see a book about core Unix programming on a Mac still has me to a doubletake the same if I saw a book called "A vegetarians guide to cooking steak."

Does Cocoa (and this book) relate to GNUStep?

I know that GNUStep is implemented in Obj-C, but would this book be useful for learning Obj-C and the GNUStep interface? Would there be any advantages for doing so (would I gain any cross-platform Linux-Mac OS X benefits)?

Carbon or Cocoa?

What should one use? I've heard that Cocoa is the way to go, but that involves learning Objective-C.. Would Carbon be supported in the future and would we be able to use all the cool features of OS X with Carbon (like Quartz)?

Aaron is an exceptional Instructor since at NeXT..

But no fucking Programming book is worthy of nearly $100 no matter the author(s).

The pass the R&D efforts of teaching Consultants how to program cost onto a potential tens of thousands of general programmer audience is the same asinine attitude some folks felt and I disagreed with while at NeXT.

Miraculously Steve thought $50k for WebObjects was steep and dropped it to $699. Amazingly, the product still exists.

Drop the price guys, get a better publisher to distribute and cover your costs so you can give us this sure to be excellent resource.

Hell, W. Richard Stevens, Advanced Programming in the UNI* Environment is definitely cheaper than this worthy book and it already has a place in history.

Good book. Good review

This is indeed a useful book. As someone who is a long time Unix programmer and just starting out with Objective C and Cocoa I found it mostly above my head but a book I know I will grow into and find useful for a long time.

I agree with the reviewer about the lack of pointers for further info. I hope that the authors might fix this at the excellent site for the book.

For close to $100 this is an expensive book, but if you want to learn to write professional Mac apps then it's probably worth the price. These guys know their stuff and the book is a lot less than the $3500 they charge for the course.

Tony Williams

Re:Paragraph Intros

No, we won't mod you down. We'll tell you. You are nit picking. All it says about his writing qualitiy is, that he was nice enough to review the book. This is slashdot. It's relaxed and depends on contributions from people.. a lot.

Unless you have something nice to say, say nothign at all. And unless he made a flagrant error, I suggest just commenting on the book. /trogdor

Re:Paragraph Intros

If you can "read" a EULA, I'm sure this is cake.

I understood is point, I didn't say "huh", nor did I otherwise get confused. Sheesh. Even the "Hacker Exploit Papers" have jokes and innuendos spiced about. And they're considered 'professional'.

Re:weird

This got a funny mod? Um, flamebit?

When you buy an Apple you get what you pay for, so that's a bad example anyway.

Who mods this stuff?

Re:weird

For example, chapter 8, "Debugging With GDB," was not simply a repeat of the online help...

Wow, so we didn't even RTFS (summary), did we? ;)

man Pages

On a side note, I have wanting to see a program developed for the *nix desktop that allows a user to select from all available "man" pages and display them in the active window. This would minimize some time I spend on the command line - not that I mind the command line, but for document viewing, editing, and note-making I would rather have a graphical interface. Let's call it "RTFM" ;) . This may be able to be done through another program and I may just be unfamiliar with its operation. Anyone???

Re:man Pages

You mean something like: ManOpen [versiontracker.com]

Re:Paragraph Intros

I don't understand. Are you expecting that the book reviewers be held to a higher standard than the editors of /.?

Re:weird

Uh, did you even read the summary?

First off, glibc doesn't have a man page. The book isn't $60, it's like $98 and even discounted it's going to be around $75-$80. Secondly, the guy specifically states that the GDB stuff isn't a regurgitation of the man page, which I assume would hold true for other pieces of the book as well.

Secondly, although I have publically stated my fundamental disagreements with Apple about their policies on patents and their general disregard for some of the fundamental concepts behind open source software, Apple makes *great* hardware -- it's *much* better stuff than you can find in value-priced x86 machines. A little overpriced, yes, but you expect that from a strong brand like Apple.

Re:important question:

it says in the Acknowedgements:
"...All layout for this book was done using OpenJade, TeX, and Norm Walsh's DSSSL stylesheets for the DocBook DTD."

In other words....

What I consider critical improvements, you consider "proprietary/design flaws compared to just Unix".

Re:mac problem

Yeah, I suppose you are running LocalTalk over a 56k modem or something and running a broken System 7 installation?

Try to copy the 17 Meg file with your fancy 486 running Windows 95 if that suits you better, but please stop whining.