The Theory of Parsing, Translation, and Compiling: Vol. 2 (Prentice-Hall Series in Automatic Computation) (Hardcover)

by Alfred V. Aho (Author), Jeffrey D. Ullman (Author)

Product Details

• Hardcover: 460 pages
• Publisher: Prentice-Hall (January 1, 1973)
• ISBN-10: 0139145648
• ISBN-13: 978-0139145643
• Average Customer Review: 5.0 out of 5 stars See all reviews (2 customer reviews)
• Amazon.com Sales Rank: #303,548 in Books (See Bestsellers in Books)

Customer Reviews

Most Helpful Customer Reviews

10 of 10 people found the following review helpful:

5.0 out of 5 stars A real tour de force - brilliantly written..., March 27, 2003

By	Mark Grindell "Mark Grindell" (Shipley,West Yorkshire) - See all my reviews

SIX stars at least. And this book, of all things, was my introduction to discrete maths - what a blast!

Here is the unsullied theory of formal languages - perhaps you should read Chomsky for a week or two just before embarking on this. All is explained though. The way that null productions make parsing so difficult is explained - the treatment of grammars as formal mathematical objects is shown via their depiction as n-tuple objects, how most computer languages are LL(1) rther than LL(2) (which I think ADA is rumoured as being), parsers as similarly structured n-tuples with a structure in some way a conjugate of the grammar....

And so forth. Wonderful!

I started with this and was amazed to discover the existence of a much simpler book later on which had a more "engineering" writing style.

This, being more theoretical but probably far more powerful in the end, is wonderful for people resarching into not just compiler design, but looking at natural language theory.

Frankly, the subject matter is well covered, but a few things might be considered additional to the text.I'mnot going to say "missing", that would not be fair.

Firstly, you would have to go to other sources to find the way in which attribute grammars have become important. These combine evaluation and parsing structures in a most elegant way, and are exemplified by languages such as ALADIN for parsing and compiling ADA. You won't find much of that sort of thing here, but it will doubtlesly help.

Secondly, the jump to full denotational semantics is not made at all. There are vague references to semi-infinite domains somewhere towards the end, but the theory is never really discussed in any way that would give you access to the sorts of semantics tools that lead eventually into full category theory, which is where I believe the real action is.

The next thing for you to read, after begging, stealing,or (heaven forbid, buying) this unabashed masterpiece is to read the seminal papers by Thatcher, Goguen, Wright, and so forth, from the 80's IBM research releases and go from there. You might also try talking to the publications dept at the Oxford Programming Research Group, Mike Spivey in particular...

The real power behind this book is the entry it affords to extensive correctness proof techniques. I believe a group originally from Oxford were working on this kind of thing using the Aho formalisms using Edinburgh LCF.

My word, even if your not in this game, you should get this book, its very approachable, and it is a real treat. I wish there were more texts like this around, maybe parallel to this showing the next stages, but I've never seen them.

1 of 2 people found the following review helpful:

5.0 out of 5 stars Computer Science, September 26, 2008

By	Carlos Jorge (Lisboa, Portugal) - See all my reviews

The book, Theory of Parsing, Translation and Compiling, by Alfred V. Aho, and Jeffrey D. Ullman, is intended for a senior or graduate course in compiling theory. It is a theoretical treatment of a practical computer science subject. Since computer science is an ever changing area of study, this book emphasizes ideas, rather than specific application details. The algorithms and concepts presented in the book should survive to new generations of computer technology, programs and systems. Numerous examples are given, with specific context, rather than on the large complicated contexts normally found in implementations, even in cases where the theoretical ideas are difficult to understand in isolation.