Lambda the Ultimate

... is that a fair amount of effort was put into slimming down EOPL in the second edition. My hardcover 2nd ed. is approximately half as thick (!) as my hardcover 1st. The chapter on the CPS transformation was essentially totally rewritten, for the better, as far as elegance and performance of the transformation is concerned.

Some reviewers claim that some of the more advanced information in several chapters was altered or removed. It's tempting to say that this must be the case based solely on the difference in size of the two books, but I'm reluctant to do so without literally going through them side-by-side ("Issues of typography alone may make the comparison an unfair one," as Guy Steele said about comparing the sizes of the Scheme and Common Lisp standards). IIRC, Ehud offered his opinion shortly after the 2nd ed.'s publication that while 2nd ed. might be better in a classroom setting, he preferred the 1st ed. for self-study.

As I write this, I feel mildly guilty for seeming to answer all of falcon's questions with lots of recommendations for textbooks whose price tags cause my wife to glare at me when she sees the Amazon bill—and I'm 20+ years beyond my starving student days. So I'm now inclined to say that there's nothing (modulo errata, of course) wrong with the 1st ed. at all. Grab it and DrScheme's EOPL language and enjoy!

In fact, I prefer EOPL1's style, but since OO and type systems are important subjects these days, I chose to teach from EOPL2.

I have read through the 1Ed before the publication of 2ed. And later I skimed the 2ed. My conclusion is that the 1ed is better.

I'm curious if folks know of other books that they feel went downhill in later editions, or where the the newer edition was a mixed-bag (as EOPL seems to be).

I can think of two examples off hand: the Green Dragon vs. Red Dragon books (on compilers); and "Introduction to Automata Theory, Languages, and Computation" (2nd ed.).