Errata for The C++ Standard Library From Scratch

Last updated 05/19/2003

Introduction

1. Page 7, under What Tools Do I Need?, first paragraph, last sentence: The exercises are not scattered throughout the book but are available on the support web site.
2. Page 8, first paragraph (continued from previous page) describes how to obtain the egcs 1.1.2 compiler. The egcs compiler has been improved and re-integrated with the Gnu compiler and is now known as the Gnu Compiler Collection (GCC) 2.95. You should probably use GCC 3.3 instead of egcs 1.1.2. However, as of 19 May 2003, the examples and exercises have not been tested with GCC 3.3. Many work-arounds are no longer needed. In particular, the <sstream> work-around included in the example sources is not needed with with GCC 3.3. See, also, the README.txt file that comes with the code.
3. Did I mention that you need to go to the Dinkumware web site, http://www.dinkumware.com/vc_fixes.html if you use VC++ 6.0? Read the big warning text in orange above.
4. The correct email address to reach me is: stdlib-scratch@halpernwightsoftware.com

Chapter 1

1. Page 18, Figure 1.2: The arrow heads pointing to Address and Appointment should have asterisks next to them, indicating a one-to-many aggregation. A portion of this diagram is drawn correctly in Figure 3.1, page 63. Thanks to John Bell for pointing out this omission.
2. Page 18 clarification: In the second-to-last paragraph, 5th sentence, "The top section is the class name, ..." would be clearer if it said "The top section of each box contains the class name, ...". That is, top, middle, and bottom refer to each box, not to the whole diagram. Furthermore I should point out that what looks like double lines in Figure 1.2 are actually two separate lines bracketing an empty section in the box. Thanks to Mark Chariton for pointing out this lack of clarity.

Chapter 2

1. Page 38, Listing 2.6, line 9: the word "The" will actually show up as "eThe" if you run the program. A copy editor fixed the "spelling error" at the last minute, not realizing that it was supposed to be that way.
2. Page 44, Listing 2.10, line 6 should read "using std::strcpy; using std::strlen;"
3. Page 55, second-to-last sentence. "Listing 2.17 is an attempt to do the same thing as Listing 2.15, ..." should read "... do the same thing as Listing 2.16, ...".  Thanks to Mark Chariton

Chapter 3

1. Page 65, Listing 3.4, Line 1: The file name is AddressBook.h, not AddressBook.cpp
2. Page 68, Figure 3.2: In the left-middle part of the diagram, the words tmplt<int> should be mytmplt<int>.
3. Page 68, Listing 3.6: Line 1: The file name is missing. It should say "File AddressBook.cpp"
4. Page 68, Listing 3.6: Line 3: #ifndef should be #ifdef

Chapter 4

1. Pages 86-88, Listings 4.5 - 4.8: Every occurance of the word, addrlist, in these listings should be replaced by std::list<Address>. The addrlist typedef is not introduced until Listing 4.9. Thanks to RTMOUNT.
2. Page 99, Listing 4.6 (continued from previous page), line 67 should be a blank line (except for the line number). The word "output" was accidentally inserted during production.

Chapter 5

1. Microsoft Visual C++ users take note: If the input to the editor always seems to be a line behind (i.e., nothing happens until you press return twice), then you have a version of the VC++ library with a number of known bugs. I mentioned this in the introduction, but many people seemed to have missed it. Dinkumware, Ltd., the company that supplies the library to Microsoft, has fixed many of these bugs. A list of fixes is available on the Internet at http://www.dinkumware.com/vc_fixes.html. You can also purchase a complete, updated library through this Web site.

Chapter 6

1. Page 157, Figure 6.2, The first column heading should read addrById_, not addrByName_.  Thanks to Mark Chariton

Chapter 7

1. Page 172, Listing 7.1 (continued), line 69, the word cachedLast_ should be changed to cachedFirst_ in the comment. The same change should be made on page 178, Listing 7.3, line 71. Thanks to Darrin Roberts.
2. Page 196, Listing 7.9 (continued), Line 127:

127:       while (iter != endIter)

Should be replaced with:

127:       while (*iter < *endIter) // empty loop if *endIter absent

This change prevents a potential endless loop if the user edits a record such that it is no longer in the search results. This change will not prevent the item from showing up in the list, but it will prevent the endless loop.

Chapter 8

1. Page 207, The second to last sentence in the Excursion should begin "We call pop ...," not "We call top..."
2. Page 208, Listing 8.3, line 6, __GCC__ should be replaced with __GNUC__. Same change for page 233, Listing 8.17, line 12.
3. Page 225, first paragraph, second line, "0ñRAND_MAX" should read "0 - RAND_MAX". Thanks to Kevin Seghetti
4. Page 226, in the excursion, first paragraph, fourth line, "ostringstream, ifstream and istrstream" should read, "istringstream, ifstream and istrstream". Thanks to Kevin Seghetti
5. Page 232, at the end of the excursion, the code given to remove and delete all the pointers in a container will not compile. The correct code is as follows:

  while (! myset.empty())
  {
    myclass* p = *myset.begin();
    myset.erase(myset.begin());
    delete p;
  }

If the container type is not known or is known to be vector, then the following is more efficient:

template <class Container>
void deleteAndClear(Container& mycontainer)
{
  // This code works for all standard containers except map and multimap
  while (! mycontainer.empty())
  {
    Container::iterator i = mycontainer.end();
    Container::value_type p = *--i;
    mycontainer.erase(i);
    delete p;
  }
}

The function template above works only for containers with bi-directional or random-access iteratators (including all of the standard sequence containers and the set container). Neither of the above code fragments work unchanged for map or multimap containers, although the same basic techniques can be used for maps or multimaps container pointer keys and/or pointer values.