This is not exactly a review of Yannis Haralambous’ Fonts & Encodings; that would be the work of years, and I doubt there’s anyone in the world qualified to discuss the whole thing, except its author. This new O’Reilly book is about a thousand pages in length. It’s impossibly ambitious, irritatingly flawed, and probably only comprehensible to a single-digit number of thousands of people world-wide; but for those people it’s an essential book, you just have to have it.
Language Shift · I was most of the way through the 25-page Introduction, not having looked at the front or back cover very carefully, and I thought “This is translated from French.” True enough, and fairly flowery French at that; for example:
For computerized typesetting is based on mechanical typesetting, and the terms that we use today were invented by those people whose hands were indelibly blackened, not with oil (the liquid that pollutes our ecosystem), but with printer’s ink (the liquid that bears wisdom).
I spotted one little probably-in-translation problem: the discussion of XML uses the term namespace alias rather than prefix. Perhaps there are others lurking in other arcane technical corners. Still, this doesn’t seem serious.
Scope · Like the title says, more or less everything having anything to do with digital text, its encoding, transmission, and presentation. Most subjects are accompanied by exhaustive historical background, which certainly make the book more enjoyable even if they don’t increase its reference value.
Let’s see: Pre-Unicode; Unicode; Asian texts; Font management on Mac, Windows, X Windows, in TEX, and on the Web; Classification of typefaces and fonts; Editing and creating fonts; Optimizing rasterization; Advanced Typography. Then there are huge, immensely detailed appendices on font formats ranging from TEX to TrueType, OpenType, and AAT. Some of these things I only barely knew existed, now I have all the details right here on the bookshelf.
Organization · The organization of this book is, frankly, weird. The discussion of Unicode starts with the encoding forms before it gets into the character repertoire and planes. The discussion of XML appears as a subset of the discussion of SVG. But the index seems competent, and as I said, more or less everything appears somewhere or another, so this is bearable.
Praise · The discussion of Unicode is pretty wonderful; the only instance I’ve seen that actually improves on the Unicode spec itself. I learned all sorts of things I hadn’t known, and I’d thought of myself as fairly erudite on the subject.
The discussion of font and typeface classification dives deep on a subject I hadn’t even known existed. Most people don’t know what Panose is, and most people who do, avoid thinking about it; the explanation here is clear and straightforward.
But really, the best thing about the book is is breadth; it’s all here, and it’s nice to have a book with all this stuff, even if it it isn’t all perfect.
Irritants · It’s not all perfect. I can only criticize the parts I can understand, but it’s not hard to find gripes. The discussion of XML is cursory at best, omitting the fact that its characters are by definition Unicode, or any discussion of Draconian error-handling; these are two things that any serious user of XML Needs To Know.
The discussion of Web typography seems pretty weak to me, too. To start with, it states, and then continues to assume, that CSS really only applies to XHTML, which is simply wrong and dangerously misleading. Then it invests a lot of space in various CSS properties which, it notes, are not actually implemented by any browser.
Then there’s a lengthy discussion of font downloading, which in my experience still lies way outside the mainstream, ending in a lengthy paean to something called GlyphGate as the One True Ultimate Path to good Web Typography. Apparently it’s a proprietary server extension that does some magic to enrich typography in your browser. I’d never heard of it, and I’m unimpressed by the Web site.
Some might find it irritating that the book invests hundreds of pages in subjects and technologies (many TEX-related) that are really only of historical interest, but I don’t see any harm in that.
Why Does It Matter? · Go check out Rob Sayre’s recent Firefox 3 Can Render Technology Correctly, which has a screenshot of a browser engaging in [gasp!] kerning. Making the browser do this involves applying a combination of technologies that are described in several hundred of this book’s pages, and, as far as I know, not in one place anywhere else. Speaking as one who invested some hours hand-kerning the title over at Textuality, I appreciate why this book needs to exist.