O'Reilly Radar has a roundup of the popular programming books (via Alex Barnett). They consider Rexx an irrelevant programming language. Aww! I actually still use Rexx on my Palm Pilot (I wrote a Twitter client in Rexx for Palm). According to this report, ASP is losing ground compared to Rails. I'd like some analysis of how different technologies within the Web stack are doing. For instance, it'd be interesting to see how the different XML-based technologies are doing comparatively. 
I'd also be careful about drawing any 'I shouldn't learn this' type of analysis from this. Book purchases correlate with how hard the language is to learn and how fashionable it is and how bad the online or in-built documentation is. For instance, I have bought Eric van der Vlist's book on XML Schema, but I don't use XML Schema as much as RELAX NG. XSD is a pig of a schema language, while RNG is great.
On SemWeb topics, I have difficulty suggesting books for people to read. Practical RDF (Shelley Powers) is one I do recommend, although beyond that, there's nothing really worth reading (there's a crapload of books titled 'Semantic Web' which have nothing to do with the Semantic Web too, which hardly helps us help other people understand what the Semantic Web is). That's part of the motivation for GetSemantic.
I haven't really bought any books on Python, because it's very well-documented internally and on the Web (Dive into Python, for instance). The same is true for UserTalk - O'Reilly do publish a book for it, but there's decent online resources (DocServer, some built-in docs for various functions and the full text of Matt Neuberg's book). Ruby, on the other hand, most people buy PickAxe and a Rails book.
O'Reilly's statistics are interesting, but they aren't everything when picking what to learn. Don't trust me either. I never pick the trendy technologies (UserTalk, RDF - I bet Fortran-powered coffee machines are next).
