Re: Timeless Classics of Software Engineering

Tony Nelson <tonynlsn_at_shore.net> wrote in message news:<tonynlsn-CE01B6.14383930072004_at_news.primus.ca>...
> In article <b2e4b04.0407300627.392f0a20_at_posting.google.com>,
> jcoffin_at_taeus.com (Jerry Coffin) wrote:
>
> > ...Likewise, the Dragon
> > Book could be used as a study in SE with compilers as the sample code.
>
> That book is far more popular than it deserves to be. The authors have
> a positive talent for making obscure exposition seem meaningful, without
> covering the necessary material, while making what they cover more
> complicated than it really is.

Well, the dragon book isn't bad, but I think a compiler book which covers more recent developments would be needed.

I think compilers have changed a lot since 70s. For instance, the book doesn't cover how to design a compiler for a parallel programming language (not that we know how to, but it should be emphasized that program representation must contain information suited to our goals). There are also a lot of parser generators nowadays that go beyond what is discussed in the book. If anybody wants to write a C++ parser, the book will be only marginally useful... I'm also thinking of the wonderful combinatorial categorial parsers and the like, these can be used for artificial languages. For instance, I've found the Parsec parser library for Haskell to be a wonderful programming exercise.

Best Regards,

--
Eray Ozkural