Most readers of books and articles about music are probably more interested in context and biography and in an impressionistic rather than deeply technical approach to describing the music itself, perhaps focused on the "meaning" of a work in extra-musical terms. (Interestingly, it's writing about popular musics that shies away most from technical description.) But — and perhaps it's the bias of a practicing musician, and a composer at that — I honestly prefer more technical descriptions and am eager to have some better ideas about how a piece was put together. I'm not embarrassed at all about discussions of materials and systems and processes and plans. I like lists and counts and charts and diagrams and lots of notated examples. And however provisional the relevance of a particular method may be to the madness of the music which ensues, I'm all in for the ride, because my intention is to learn, copy (which is composers' polite-talk for steal) and adapt, whenever something, even the tiniest trick, appears promising.
In connection with some recent deep study of music by Berlioz, a figure I consider as important and as radical as Ives and Cage, Julian Rushton's The Musical Language of Berlioz has been a constant companion. While Berlioz's life and times provides (and has often provided) ample material for biographers and cultural historians, the real interest is ultimately in the music itself and getting a handle on that achievement simply requires talking about technique, which Rushton does beautifully and concretely. More importantly, for me as a composer, Rushton lays out the material elements of Berlioz's practice in such a way that they are constantly suggestive of paths for making music that mainstream music history more or less abandoned, cul de sacs well worth re-exploring.
Books like Rushton's might be described as constructive, and indeed they read a bit like do-it-yourself books for building this or that (as a kid I had a small collection of stage magic and puppetry and outdoorsy arts-and-crafts books, so it's long been one of my favored genres.) Lou Harrison's Music Primer and Messiaen's The Technique of My Musical Language are good examples of a composers letting us in on how they work in a direct and constructive way and I've been lucky to have learned from both since High School. There are also useful bits by Boulez and Reich and Stockhausen writing as concretely about their own music (Stockhausen's little book on In Freundschaft is practically a recipe for making the piece.) Ives' Memos have some wonderful hints about technique as well. While much maligned elsewhere, I have long been enthusiastic about John Cage's analytic half to the the Cage/Hoover Virgil Thomson; Cage has a refreshingly direct way about discussing a piece of music, unafraid of counting notes and often salvaging useful technique from a piece that has otherwise sunk. (One of the greatest hours in my life was spent in a hotel in Houston when Cage discussed his understanding of Mozart's materials and methods; I would gladly loose a limb for a time machine and a tape recorder to recover that lesson!) Charles Shere's elegant book on Robert Erickson, Thinking Sound Music, remarkably conveys a lot of usefully concrete information almost exclusively without the use of notation. Finally, I find three examples from ethnomusicology richly suggestive to composers: Paul Berliner on the Mbira, Michael Tenzer on the Balinese Kebyar orchestral style, and Simha Arom on African polyphony.
(Here's an older item with my list of favorite books on composition.)