Errata

Errors in the Second Printing

Chapter 4

Figure 4.14, step 2(d): "1 + \log_e (N/f_t)" should be "\log_e (1 + N/f_t)"; and
Figure 4.14, step 2(f)ii: "\log_e (1+f_{d,t})" should be "1 + \log_e f_{d,t}";
(Marcelo Mydlarz, Rutgers University, March 2000).

Additional Errors in the First Printing

Preface

Line 18: "catch phase" should be "catch phrase".
(Chris Manning, Stanford University, November 1999).

Chapter 2

Line 9: "proprtional" should be "proportional".
(Yugo Kartono Isal, University of Melbourne, October 1999).

Line 10: "contructed" should be "constructed".
(Yugo Kartono Isal, University of Melbourne, October 1999).

Line 9: "The characters about to be encoded" should be "The characters about to be decoded".
(Ellen Spertus, January 2005).

Table 2.7: the font used for the file "HTML" should be sans-serif italic like the other file names, not small caps.
(Tim A.H. Bell, University of Melbourne, July 1999).

Chapter 3

Line -7: the LaTeX math-mode expression

    x = 1 + \lfloor (\log (1-p))/(\log(1-v))\rfloor

is incorrect, and should be

    x = 1 + \lfloor (\log (1-v))/(\log(1-p))\rfloor

that is, the numerator and the denominator of the fractional part need to be exchanged.
(Mike Liddell, University of Melbourne, July 1999).

Line 12: the value b=2,036 listed here should be b=2,039, as calculated on the previous page.
(Penny Liang, Tsinghua University, March 2008).

Chapter 4

Section 4.8, paragraph 2, line 4/5: "Yuwona and Lee" should be "Yuwono and Lee".
(Yugo Kartono Isal, University of Melbourne, October 1999).

Line -4: "Yuwona and Lee" should be "Yuwono and Lee".
(Yugo Kartono Isal, University of Melbourne, October 1999).

Chapter 5

The third line of the displayed mathematics is in error, and should be

    10 f t_r + 2 I' t_d + I (t_r + t_d)

that is, add a multiplier of t_r to the first term.
(Vicky Wang, Queens College, City University of New York, October 1999).

Chapter 6

Line 2: "For an eight-bit grayscale image, there are only 511 possibilities" should be "For an eight-bit grayscale image, there are only 255 possibilities".
(Glen Gibb, University of Melbourne, May 1999).

Chapter 7

Figure 7.8: There is a missing pixel in the top-left section of the six parts of the figure: just below the stroke of the "e" there is a pixel that should be black, namely, the pixel corresponding to the one labelled "1" under the stroke in the top-right section of the figure.

Separate errors exists in the top-right section of the same figure: the label "4" on the first pixel in the second row should be "3", and the labels "3[gray]32[white]" on the first five pixels in the third row should be "3[gray][gray]22".

As a consequential change, the last sentence of paragraph 1 on page 327 should be "The total weight is 73 for the two es, but 131 for the c and o".
(Glen Gibb, University of Melbourne, June 1999).

Figure 7.9: The same pixel labelling errors in the top-right section of this figure too: the "4[gray][gray]..." in the second row should again be "3[gray][gray]...", and the "3[gray]32[white]" in the third row should be "3[gray][gray]22"

As a consequential change, the last sentence of paragraph 1 on page 328 should be "in fact the total error weight between the es in Figure 7.9 is 129, almost the same as that between the c and the o in Figure 7.8".
(Glen Gibb, University of Melbourne, June 1999).

Chapter 9

An incorrect impression is created by this section that suggests that the Larmore and Hirschberg (1990) description of the package-merge process requires O(nL) space. A straightforward implementation of their method does need this much space, but in their paper Larmore and Hirschberg also show how to reduce the space requirement to O(n), by not retaining intermediate values in the computation, and performing a controlled amount of re-evaluation.
(Guenter Rote, Frie University of Berlin, June 2003).

Appendix A

Para 2, line -1: In fact, we still do not have a version of mg that operates correctly on DEC/Compaq Alpha hardware, sorry. Anyone out there want to take responsibility for the port?

Section A.1, line 2: The sole repository for the MG source code is now the MG web site at http://people.eng.unimelb.edu.au/ammoffat/mg/; the ftp site on munnari.oz.au is no longer valid.

Lines 4, 5, and 6: These three lines have been a bit mangled, and should be "the location where the file is to be stored. For example, the command to compress the file monalisa.pgm with mgfelics is"
(Owen de Kretser, University of Melbourne, August 1999).

References

Line 16: "Yuwona, B., and D.L. Lee. 1997" should be "Yuwono, B., and D.L. Lee. 1997".
(Yugo Kartono Isal, University of Melbourne, October 1999).