Typesetters are not machines
I tell you three times:
Typesetting is not conversion.
Typesetting is not conversion.
Typesetting is not conversion.
A plea hit my inbox today for help with a so-called “mass conversion” of SGML and non-marked-up ASCII text to PDF.
I turned it down. Firmly. And I could use the money, what with some thirteen or fourteen thousand dollars of library-school tuition staring me in the face.
I am, however, an independent now, and I don’t have to pretend to go along with the utterly ludicrous. TYPESETTING IS NOT CONVERSION, people! If it were, every typesetter in existence would have been downsized long ago in favor of programmers running batch-pagination systems. Ain’t happened.
The problem, as usual, is the page. Computers are perfectly capable of plonking text down in a randomly-sized block; what do you think your web browser is doing right this very moment? Unfortunately, they pretty much suck at it, as witness any page whose designer has the temerity to use full justification. They don’t hyphenate right, especially anything outside their private dictionaries. They allow horribly tight or loose lines. They let the last line of a paragraph have a piddling two or three characters.
When the block is bounded at the top and bottom as well, as a book page is and a web page isn’t, even more horrible things can happen. Widows. Orphans. Heads abandoned at the bottom of a page. Words hyphenated across pages.
Now, incredibly sophisticated batch-pagination systems permit setting parameters in an attempt to control these problems. Does that mean no human decision-making after setup? (Not that setting one of these puppies up is exactly a small task.)
Nah. I used to jockey such a system myself. It was good—often amazingly good—but it got itself into dilemmas it couldn’t solve (“I can’t put three lines of text below this head! Help!”). My job was getting it out of them, and fixing the things it did wrong. (Hyphenating romanized Japanese words. That was a killer. I did this book on Japanese history, you see… corrected more cases of “Yamam-oto” than I care to recall.)
Yeah. I used to cuss the thing. A lot.
Point being, there is no way to treat typesetting as a completely mechanized process (which is what “mass conversion” implies) unless you can tolerate less-than-perfect, web-page-ish results. OEBPS-based electronic books make that trade: typesetting perfection for pagination and text-size flexibility. As I’ve said before, I have no trouble with the trade; the gain vastly overshadows the loss.
With a PDF, however, the output of a mechanized conversion will be compared to humanly-typeset documents. Unfavorably.
I did the right thing. I really did. I told them about FOP and XEP and PassiveTEX. I also told them I don’t know how to use these things and they don’t want to pay for my learning curve.
I foresee some screaming matches in the future, as they learn the hard way what I said three times at the beginning of this post. But when the screaming dies down, perhaps they’ll have learned something.
Go hug a typesetter today, won’t you?