Customer Reviews

Write Great Code, Volume 2: Thinking Low-Level, Writing High-Level

5 star:		(6)
4 star:		(1)
3 star:		(0)
2 star:		(1)
1 star:		(0)

 

 

I earned my Computer Science degree several years ago. During my studies I learned languages like Assembly, Modula-2, C and C++ and even a little VB in a computer graphics class. I learned quite a bit about how computers work at the binary level and even got into some Electrical Engineering topics.

As you might expect, I don't use any of those languages in my career today. It's all Java, LotusScript, and other high-level languages for me these days. I rarely find myself shifting bits or thinking about logic gates. Does this mean my CS degree was a waste of time? Absolutely not. Although it would have been nicer if I had been learning Java... :-)

It's not the languages I learned that gave the education its value. It's the algorithms, patterns and logic that have come to my aid again and again. An understanding of what's going on inside the computer can be very helpful at times. That's why I was so interested in reading Randall Hyde's book: Write Great Code, Volume 2: Thinking Low-Level, Writing High-Level.

This book teaches the following concepts (Chapter 1 excerpt):
- Why it's important to consider the low-level execution of your high-level programs
- How compilers generate machine code from high-level language (HLL) statements
- How compilers represent various data types using low-level, primitive, data types
- How to write your HLL code to help the compiler produce better machine code
- How to take advantage of a compiler's optimization facilities
- How to "think" in assembly language (low-level terms) while writing HLL code

The general goal of this book is to teach you how to think like a compiler so you can anticipate what the compiler will do with your code. Randall Hyde is also the author of The Art of Assembly Language. It would be an understatement to say he has a good handle on this subject. Chapter 3 goes into detail on 80x86 assembly for the HLL programmer while chapter 4 looks at the Power PC. Chapter 5 looks at compiler operation and code generation. This chapter gave me major flashbacks to compiler class.

In chapter 9, the author includes a great discussion of arrays and how they are handled by various languages. Another topic, in chapter 12, that I found very interesting was the discussion of variant data types. (Us LotusScript developers mustn't forget to use 'Option Declare.') Chapter 15 includes a description of various types of loops and hammers home the importance of coding efficient loops.

We spend so much time these days trying to abstract everything and think at a high-level. That's all fine and good but we can make better high-level decisions by understanding the low-level workings of compilers. Readers who are experienced developers with a genuine interest in writing quality code will get a lot out of this book. I especially recommend it for career developers who didn't have the benefit of formal Computer Science training.

At the end of the book is a teaser for the next book in the series: Write Great Code, Volume 3: Engineering Software. In this book, Hyde will discuss "personal software engineering" and will focus on craftsmanship, art, and pride in workmanship. I'm looking forward to that one too.

Back in "the day", you really couldn't write high-level code without at least some exposure at some point to lower-level code, like Assembler. Now, you can pretty much be completely ignorant of what happens in your Java or VB code "under the covers". But that doesn't mean you can't benefit from understanding how your compiler turns your readable code into machine-readable operations. Randall Hyde does an excellent job in explaining all this in his book Write Great Code Volume 2: Thinking Low-Level, Writing High-Level.

Contents: Thinking Low-Level, Writing High-Level; Shouldn't You Learn Assembly Language?; 80x86 Assembly for the HLL Programmer; PowerPC Assembly for the HLL Programmer; Compiler Operations and Code Generation; Tools for Analyzing Compiler Output; Constants and High-Level Languages; Variables in a High-Level Language; Array Data Types; String Data Types; Pointer Data Types; Record, Union, and Class Data Types; Arithmetic and Logical Expressions; Control Structures and Programmatic Decisions; Iterative Control Structures; Functions and Procedures; Engineering Software; A Brief Comparison of the 80x86 and PowerPC CPU Families; Online Appendices; Index

This is the type of book that will really excite you if you're wondering why a nested if statement performs differently than a case statement. Hyde explains basic compiler theory, and applies that to how your compiler of choice decides on optimization strategies. It's impractical to get a program optimized for all factors, like code size and speed, but there are reasonable trade-offs as well as compiler options you can use to prioritize one factor over another. You also don't have to be completely conversant with Assembler in order to work through this book. He uses a number of high-level coding examples from various languages (like C, C++, and Pascal) that should be understandable to most any IT developer. He then shows the translated code at the assembler level and explains why a particular compilation might be good or bad (depending on your requirements). Once you start to understand how (and why) your compiler "optimizes" your code, you can make more intelligent choices as far as programming constructs. There is often a dozen ways to do something, but a few of those ways will be far more efficient at the machine level than the others.

If you're writing a program that runs in a second or two, these techniques may not necessarily have an immediate practical application for you. But that hour-long program may have the same underlying problem as your first program, and it could be that a slight change in program structure at your end could dramatically reduce the run time. This is one of those books that most serious software developers should have sitting on their shelves...

I read the first volume of this book, and it was a great, informative read. After volume 2, I have this to say:

This book is not a cookbook for writing better code. Hyde explains why certain programming constructs are better than others (and in what cases), and backs it up with evidence from the assembly code (that is the entire premise of the book). Finally, solid proof of (and against) what I've been hearing all along from instructors and other programmers on message boards or face-to-face communication. That in mind, it would have been nice to have a summary of the tips at the end of the chapters, or the end of the book, as a quick-reference kind of thing. These concepts are the perfect thing to consider when fine-tuning your code. I take the stance that if you fine-tune as you go, you have less work later, so I took notes as I read and have started implementing changes for the better (with evidence that it is better) in some of my coding.

Does anyone write code in Pascal anymore? Seriously? Hyde discusses examples in C/C++ and Pascal for the most part (favoring Pascal, by my estimation), so it is nice that the book is language-independent for the reader. The assembly examples in the book are in 80x86 and PowerPC...I think it would have been better to release two versions of this book, one where the assembly is 80x86 (because it is so ubiquitous, if for no other reason), and then another where the assembly is in PowerPC assembly. I didn't pick this up to become an assembly language programmer, and quite frankly, the PowerPC examples just confused me. It looks like my cat stepped on my keyboard and it appeared in print.

The tone of the text is sometimes condescending (especially in the first 5-6 chapters), with a not-so-subtle "real men only code in assembly" message. It seriously reminds me of my parents/grandparents whining about how hard they had it growing up...walking barefoot to school in the winter...and only having assembly to work with in writing programs.

So by the end of this book, I think I have met Hyde's goal for the readers. I can't sit down and write an assembly program, but I can read enough assembly to compare different versions of my high-level program to say what is better. As the subtitle suggests, I am "thinking low-level" now as I write my programs (in C...which most programmers would call low-level). I suggest "actively" reading this book and coding up a few examples with your compiler and examining the output, following along in the text. Doing that really made the material click for me.

Overall, it's a good book and definitely worth it. This is a long one though, and takes some time to digest. See you in volume 3!

How Software Works would also be a great title for this volume. Any high level construct you can think of is analyzed in-depth at a low level. This is great for those interested in reverse engineering or writing cleaner code.

This book is language and hardware neutral. Languages from Pascal to C++ to Visual Basic are covered on hardware from x86 to RISC to embedded systems. No matter what language you code in the innards are laid out for you. The author does a masterful job of the idiosyncrasies of different languages seamlessly without dwelling on the trivial. This book is heavy on assembled code, compilers, and interpreters. It's an absolutely great collection of everything you wanted to know about how software works that you really can't find elsewhere in a single volume.

Although the subject matter is probably dry for some, I found this book surprisingly easy to read due to the clarity of the writing. This is at least as good, or better than volume one. With that said, there are a number of references to volume one, and I do recommend reading volume one first unless you have a solid background in computer architecture.

In this RAD time when anybody who knows drag-and-drop can proudly claim himself software developer, this is the unique book for true, serious, and hardcore programmer that writes native codes. I completely agree that, in order to write efficient and optimized code, you not only need to understand what your compiler can do for you,but also you should know, at least for a minimum level, what you can do for you compiler. In this regard, this book does an excellent job explaining the magic your compiler does for you and what you can do to become part of the magic. As such, I highly recommend this book to any serious programmers.

The only thing I don't like about this book is that it appears to be giving too much space to HLA, which I suspect is a widely accepted assembly dialect that was actually used in any real-life project. And I just feel it is a waste of my time trying to understand and catch up with this HLA language. This results in my taking one star off with the final rate of 4 stars.

Other than that, this is indeed a very good book.

In a previous reivew, it is complained that Java is not touched. This is understandable while it really doesn't seem to be a problem to me. Java after compiling is translated into bytecode that has to be executed on the virtual machine. I guess that is really beyond the intended scope of this book, which is related to the world of native codes that after compiling can be directly executed by the machine.

The book is really good in teaching the best way to code in order to produce faster/smaller programs, taking advantage of the hardware. A lot of the concentrated info contained here is hard to find elsewhere and would require extensive research in several compiler theory and computer architecture manuals.
After reading the text your understanding of software/hardware interface will be improved by several orders of magnitude.
The only small annoyance is the fact that a lot of times Mr Hyde start the discussions on various topics using HLA listings as example and even if he provide details on MASM/TASM applications he does so comparing these to HLA. Hence you have to always study the HLA examples ( HLA is some kind of low-level language the author developed ).
Since I'm not interested in HLA ( probably like a lot of the readers out there ) I did not like the ( small ) additional effort it required.
Overall this book is a little gem.

Very few books go into the kind of detail that this one goes into. It teaches you to think like a compiler, so you understand the reasoning why certain code is slower that others. It is definitely a good book to get to learn how to optimize code. It talks about current game systems, and assembly that those would use also.

The first 100 pages of this book were fantastic for concepts. After that, there were numerous code listings of various dialects of assembler that were each several pages in length. The premise is that you should learn to read assembler so that you can write better high level code. However, there is no tutorial on what the various assembler keyword instructions mean. This book feels like it was slapped together because it was the long overdue volume two in the series, and has so many references to the first volume that you practically need to own both. There is not enough on Java (or OO in general) as the reader's high level langauge of choice, and far too much on Pascal.