Expert C Programming: Deep C Secrets 1st Edition

Peter van der Linden's "Expert C Programming: Deep C Secrets", published in 1994, is already a classic. It explicitly assumes the reader knows how to program in C (according to the author, this "should be every programmer's second book on C"). In that sense, it is similar to Scott Meyers' "Effective C++" (originally published in 1991). Of course, C is a much smaller language than C++ but, even so, it does have some non-trivial aspects, which are precisely what van der Linden zeroes in on. Given the nature of this text, any review of its good and bad points needs to get down to the nitty-gritty.

The Good: Dennis Ritchie, in his essay on "The Development of the C language", wrote that "Two ideas are most characteristic of C among languages of its class: the relationship between arrays and pointers, and the way in which declaration syntax mimics expression syntax." In the book under review, van der Linden is at his finest when discussing precisely these two topics. Starting with arrays & pointers: the book includes 3 chapters on the subject, first tackling the classic "defined as array / external declaration as pointer" problem. Later, the author returns to the root of the confusion, namely that even though arrays and pointers are distinct entities in declarations/definitions, there is one major exception: a function can have an array parameter, but the argument will be converted to a pointer before the call (though this rule isn't recursive). As a result, even though arrays are not modifiable lvalues, you can use assignment inside a function since the argument will have been converted to a pointer. Leaving declarations behind us, the story gets more elaborate: i) when used in an expression an array name is converted to a pointer, and ii) a subscript is always equivalent to an offset from a pointer. Of course this is too straightforward, so the first point does not apply when the array "is the operand of the sizeof operator or the unary & operator, or is a string literal used to initialize an array" (6.3.2.1) -- C1X adds the _Alignof operator to this list. Moving on to Ritchie's second characteristic idea, the relationship between declaration syntax and expression syntax: van der Linden has one chapter on this subject, providing a very useful algorithm (in two versions) that allows one to untangle virtually any possible declaration (though it would have been nice if he had also mentioned "go right when you can, go left when you must"). He then applies this algorithm to the signal library function (sigaction's prototype was presumably too simple). In this connection, the author makes the interesting observation that the addition of const and volatile to ANSI C led to a reduction of the number of cases in which the use of a variable mimics its declaration. In a later chapter he makes a similar point regarding calls to a function and calls to a function through a pointer (or any level of pointer indirection). These are only two examples of van der Linden's non-hagiographic attitude: he devotes an entire chapter to the aspects of C he doesn't like (C's "sins of commission, sins of omission, and sins of mission" as he calls them), bringing to mind Matthew Wilson's later volume "Imperfect C++". The author's choice of words in this case ("sins of mission") reflects his wider approach to writing: his style is relaxed and often funny (more so in the early chapters). For example, he quotes Peter Weinberger (of awk fame) as saying: "All C programs do the same thing: look at a character and do nothing with it". Finally, the book does contain a number of other insights/techniques from various aspects of C programming, some important and relatively obvious (e.g. to implement a finite-state machine in C you probably need an array of pointers to functions), others considerably less so (e.g. the typedef keyword doesn't have to appear at the start of a declaration!).

The Bad: starting with big-picture issues, it's worth pointing out that the title is somewhat misleading. The book under review is better described by its subtitle "Deep C Secrets" (pun notwithstanding), as what it mainly covers are C's dark corners. Still on the theme of what a book with this title should or shouldn't include: the middle chapters (on the runtime system and on memory) are very clear and interesting, but most of this material can already be found in any undergrad Operating Systems textbook (minus the SunOS bent). While it is true that a skilled C programmer should know this stuff, these chapters are nowhere near expert-level. On a different note, the book is showing its age. Its datedness has secondary consequences (e.g. repeated references to MS-DOS and one mention of "machines on the Internet network") but also substantive repercussions. As this book came out in 1994, it doesn't contain anything on C99, therefore no bool, inline, restrict, static and type qualifiers in parameter array declarators, or any of the other features. Interestingly, van der Linden does mention that at the time people were talking about adding a complex number type to C: this was added in C99, but it looks like it's going to become optional in C1X. Similarly, he laments the fact that gets was included in C89: it was deprecated in C99 and will be removed in C1X. As could be expected, the C++ chapter is even more temporally bound. The text also suffers from a few minor issues which ordinarily wouldn't be worth noting, but seem out of place in a detail-oriented book whose author and publisher have had 17 years to fix up lesser faults. For example, terms like "data segment" and "text segment" are thrown around in the first half of the book but are only introduced in chapter 6 (of 11). At an even more fine-grained level, the order of expression evaluation is said to be unspecified on p. 48 though it was undefined only a page earlier. Another category of minor problems has to do with issues that were mangled in later reprints despite the errata list. First, on p. 80 the early part of the section title on "typedef x int[10] and #define x int[10]" was silently changed between printings (i.e. without an entry on the errata page) to "typedef int x[10]". This raises new issues which are also not discussed in the main text: the macro expansion leads to an invalid declaration, but now the typedef is legal. Second, p. 205 used to say "the automatic promotion of a character literal causes it to become an int", but in a later printing this was changed to "character literals have type int and they get there by following the rules for promotion from type char" which is obfuscatory at best. In reality, character literals (officially called "character constants") have type int because the standard says so: "An integer character constant has type int" (6.4.4.4). [As an aside, this is one of the incompatibilities between C and C++ mentioned by van der Linden on p. 326.]. Finally, the book's index is hopeless: e.g. there's an entry on Fortran and a separate one on Fortran 90, but no entries on functions or structs.

Readers who are searching for a "second book on C" with lots of insights on design should probably look at David Hanson's "C Interfaces and Implementations", as "Expert C Programming" doesn't really discuss large-scale software construction. Also, this book does suffer from a pro-Sun/anti-Microsoft bias and it does contain some outdated material. However, a finite fraction of the C community still uses C89/C90 exclusively, so for readers who want to study not-so-often examined aspects of C this text is irreplaceable (despite minor problems here and there). All in all, 4 stars.

Alex Gezerlis

You know, it's a good book, but I didn't think it lives up to the title of being at the "Expert" level. The author suggest in the introduction, I think it was, that this should be the 2nd book for you to read on C, and that sounds about right. The "C Programming Language" to me is a work of art, and I really wish they would update it with the latest information from the standards.

At this point it's dated (no surprise), with discussions around the switch from K&R to ANSI style of declarations, and Sun specific information (about compiler switches) and authors work on Pascal is also not particular interesting. Other things are still very much relevant like how to pass arrays to a function. I like informal voice of the author, but the stories (clearly marked) rarely add any anything of value other than entertainment. There is a section of the things that C got wrong, and honestly, I don't really care, till it changes.

This is a true aficionado's delight! From the dumb pun on the front cover (PI) to the hilarious anecdotes to the meaty discussions on compiler design, this is a must-read for anyone who programs computers. Although especially pertinent to those writing in C and C-based languages (which are more than most people realise!), anyone programmer, architect, systems administrator, or designer can seriously improve their understanding of how any why our digital friends work as they do (and don't) with this book. The writing style is personable and consistently engaging, even through some of the more ethereal aspects of bit-flipping.

I have a copy on my nightstand full of annotations. This book is a joy to read. Plus, I learned something! It reads more like "Hitchhiker's Guide to the Galaxy" than it does the C&R C Language 2nd Ed. I have gifted this book to several friends.

I wish there existed many more books in the style of Peter Van Der Linden's.

C Language is brilliant and distinct from C++, each has it's strengths and weaknesses. For example, C Language has a transparency - what you read in the source is what happens at runtime.

Take this book for a test drive.

This is a book presents the strengths of C Language and where they have had the most impact. As a result of reading this book I understand C Language much better. I also understand C++ better - because I have some additional context for why C++ adds a feature, deprecates (or discourages rather) a feature.

The actual contents of the book and presentation are fairly good. It sometimes wanders a bit before making the point, but that's okay for a non-reference book on a technical subject. However, the print quality is horrible.

Not only are the pages very lightly bound (perfect binding style, not very perfect job), with pages having the top 20% pulling free of the binding the instant you open the book, but the left/right pages are different colors. The left pages have a gray / dirty / smudged look to them. The right hand pages are clean and look good. Makes it distracting to read, especially on the worst pages. I spend a lot more time trying to read what it says, instead of understanding. To bad, otherwise I'd have given it a higher rating.