by Peter H. Salus
<peter@pedant.com>

Peter H. Salus is a member of ACM, the Early English Text Society, the Trollope Society, and is a life member of the American Oriental Society. He has held no regular job in the past lustrum. He owns neither a dog nor a cat.


It's clear to me that Blake's "dark Satanic mills" must be at work printing books. The spate continued during the summer (nearly 100 publishers' announcements of Windows98 books, for example). I, however, having completed a large project (see end of column), read and browsed my way through something I first looked at 30 years ago.

Fundamental Things

There are few books that are "must reads" within the programming community. The three volumes of Knuth's Art of Computer Programming are certainly among this small band.

Since 1968, when Fundamental Algorithms appeared, Art has been revised and expanded. Over the past year, Addison-Wesley has brought out third editions of volumes 1 and 2 (Seminumerical Algorithms) and the second edition of volume three (Sorting and Searching).

The amazing thing, to me, is that these volumes, initiated before C, Scheme, ML, Haskell, C++, Smalltalk, Eiffel, etc., are still readable and relevant. Knuth may not have picked up every criticism of earlier editions, but he has clearly read them and considered whether or not to adapt his text.

Over the decades, the scope of Art has waxed. We were told by Knuth early on that his plan encompassed several more volumes. Though there are still but three, the number promised has grown to volumes 4, Combinatorial Algorithms, 5, Syntactical Algorithms, 6, The Theory of Languages, and 7, Compilers. I hope Knuth's plan is effected, mainly because I long to read volume 6.

But, as the titles reveal, as Knuth progresses, the topics become ever more specialized. However, I can recommend volume 1 to everyone: it is not a "quick read"; nor is it easy. It is a well-designed and well-written book on the nature of algorithms that are in daily use as we communicate with and through our computers.

Knuth invented TeX: these books are set in METAFONT, and a cursory tally reveals Arabic, Chinese, Devanagari, Japanese, and Korean entries in the index. An amazing piece of typesetting!

Even if you still own the first editions, you'll want these for your library; if you don't read them through, you'll want them for reference.

P.S. For some reason that has not been revealed to me, Addison-Wesley (Longman) has let Software Tools go out of print (though Software Tools in Pascal appears to be available). Even if you're not running RATFOR, Kernighan and Plauger produced a book that for 20 years embodied how programs could be made clean and easy to read, maintain and modify. It should be a must for any programmer not moving to Redmond, WA. C'mon, AWL!

In the Picture

I'm an admitted fan of Icon. So I dove into Graphics Programming in Icon as soon as it arrived (thud). Because the high-level graphics features in Icon are integrated into the language (unlike, say, C and C++, which use additional graphics libraries), graphics code is briefer. The book is very fine. The color illustrations are informative (as well as lovely). The CD-ROM is useful, containing binaries for Windows and several UNIXes. What more can I say?

Keeping it Secret

Encryption, cryptography, and security have become buzzwords on the nation's business pages, but there are few good books on the topics involved. In fact, I'm disappointed. Though the two volumes I've just looked at are well-written and instructive, neither Loshin's Personal Encryption Clearly Explained nor Stallings's Cryptography and Network Security is without flaws. For me, the biggest of these is shared: neither author mentions ssh nor scp (rcp for ssh). I learned about ssh last year from an article by John S. Quarterman (Matrix News, vol. 7, no. 2, Feb. 1997), and have found an increasing number of sites using it. I now use it in preference to Telnet. Tatu Ylonen and Timo Rinne in Finland deserve as much credit for this as Phil Zimmerman for PGP.

Loshin has another problem: pages 306-322 concern Outlook Express, and 322-324 concern Netscape Messenger. As most of you know, both of these are susceptible to wormlike viruses that attack the vulnerability of the overflow buffer. Tch, tch. Despite the excellent keynote last year by Butler Lampson (at the NT workshop), I think "Windows security" may be an oxymoron.

Loshin does do a good job on PGP and RSA SecurePC.

Stallings new edition is solid, but not complete.

And I'm still waiting for a complete book.

Digital Library

The British Library began a series of inquiries and projects concerning the "future" of the library in 1993. Several of the projects were part of the library's "Initiatives for Access Programme" and are presented in this lavishly produced book. (Carpenter, et. al, Towards the Digital Library)

A number of the essays affirm that the British Library will continue to conserve and acquire books and manuscripts. I hope so. But the general tenor of these repeated remarks is to render me pessimistic. And one wonders.

Will everything really be digitized eventually? We fret over losses of cinema, of audiotape, etc. But we know and understand the limitations of paper ­ Cotton Vitellius A.xv (Beowulf) in the library and the original Magna Carta bear ample witness to its survival power and that of parchment. But what do we know of CD-ROMs that makes us confident of their staying power?

Among the quotes from readers looking at the Sforza manuscript using the PIX program are several remarking that they could turn pages "just like a book" and see the illuminations as though they "had the book."

Really fascinating. Do we feel that these metaphors will die away? Or are the utilitarian marks on paper the standard to which all of our other methods of storage will be held?

In 1960, I obtained a "Manuscript Ticket" for the library, then in the British Museum. In 1967, finding the photographs unsatisfactory, I actually held the Beowulf manuscript in my hands and examined one page. Although the Sforza Book of Hours looks lovely on a good screen (and in this book's photographs), it does not look like the Book of Hours. I couldn't understand why for a while, but then I realized: it lacked scale. The small leaves of Beowulf and the elephant folios of Audubon are effectively equal.

We bring these works (in rendered form) to the public at large, but water them down through the democratization. ($19.95 prints of Van Gogh or Renoir or Cézanne are little like the paintings themselves; an acquaintance once complained to me in the Museum of Modern Art that she hadn't known that Dali's The Persistence of Memory was so small.

Some of the essays in Towards the Digital Library are extremely interesting. But what does it mean about the digital book that its references are shoddy and that there is no index?

Will I have to change my title to "The Digital Worm"?

Languages

I admit to a certain ambivalence where advertising myself is concerned. But the brief mention in the "Is This Issue" column in August ;login: brought me a dozen pieces of email requesting ISBNs for the Handbook of Programming Languages (Macmillan Technical Publications): I. Object-Oriented Programming Languages, ISBN 1-57870-008-6. II. Imperative Programming Languages, ISBN 1-57870-009-4. III. Little Languages and Tools, ISBN 1-57870-010-8. IV. Functional and Logic Programming Languages, ISBN 1-57870-011-6.

Books Reviewed in This Column

Donald E. Knuth
The Art of Computer Programming
Vol. 2. Seminumerical Algorithms
3rd ed. Reading, MA: Addison-Wesley, 1997.
ISBN 0-201-89684-2. Pp. xiii+762.

Donald E. Knuth
The Art of Computer Programming
Vol. 3: Sorting and Searching
2nd Ed. Reading, MA: Addison-Wesley, 1998.
ISBN 0-201-89685-0. Pp. xiii+780.

R. Griswold, C.L. Jeffery, & G.M. Townsend
Graphics Programming in Icon
San Jose, CA: Peer-to-Peer, 1998.
ISBN 1-57398-009-9. Pp. 504+CD-ROM.

Pete Loshin
Personal Encryption Clearly Explained
Chestnut Hill, MA: AP Professional, 1998.
ISBN 0-12-455837-2. Pp. 545.

William Stallings
Cryptography and Network Security
2nd. ed. of Network and Internetwork Security
Upper Saddle River, NJ: Prentice Hall, 1998.
ISBN 0-13-869017-0. Pp. 569.

L. Carpenter, S. Shaw & A. Prescott
Towards the Digital Library
London: The British Library, 1998.
ISBN 0-7123-4540-X. Pp. 256.