[tmccrmck]

We've rehashed the debate over the utility of the dragon book in the 21st century so many times on HN that I'm not going to comment on it's usefulness. (Search if interested)

But, for those interested, I highly recommend Engineering a Compiler by Cooper and Torczon OR Modern Compiler Implementation in ML by Appel. I've read all three and would only recommend the dragon book to someone very interested in the subject.

[CalChris]

The Monica Lam sections in the second edition of the Dragon Book are out of this world good. Also, she's an excellent lecturer.

Muchnick's book is still quite good.

A second semester compiler class is rarely taught and so everyone tries to squeeze too much into one semester. I almost wish they broke it in two and allowed students to take either half. LL(1) parsing isn't a prerequisite to static single assignment.

[pbiggar]

+1 for Cooper/Torczon (as someone who has a PhD in compilers, and still hates the Dragon book).

[timthorn]

Appel's book is also available in many other languages, if the thought of ML is offputting.

[sigjuice]

Why would the thought of ML be off-putting?

[timthorn]

If you're unfamiliar with it but want to learn about compilers, a text in a language you know will decrease the cognitive load.