Effective STL:
50 Specific Ways to Improve Your Use of the Standard Template Library

• Advice on choosing among standard STL containers (like vector and list), nonstandard STL containers (like hash_set and hash_map), and non-STL containers (like bitset).
  
  
• Techniques to maximize the efficiency of the STL and the programs that use it.
  
  
• Insights into the behavior of iterators, function objects, and allocators, including things you should not do.
  
  
• Guidance for the proper use of algorithms and member functions whose names are the same (e.g., find), but whose actions differ in subtle (but important) ways.
  
  
• Discussions of potential portability problems, including straightforward ways to avoid them.

—Herb Sutter, independent consultant and secretary of the ISO/ANSI C++ standards committee

—Thomas Becker, Senior Software Engineer, Zephyr Associates, Inc., and columnist, C/C++ Users Journal

• Advice on choosing among standard STL containers (like vector and list), nonstandard STL containers (like hash_set and hash_map), and non-STL containers (like bitset).
• Techniques to maximize the efficiency of the STL and the programs that use it.
• Insights into the behavior of iterators, function objects, and allocators, including things you should not do.
• Guidance for the proper useof algorithms and member functions whose names are the same (e.g., find), but whose actions differ in subtle (but important) ways.
• Discussions of potential portability problems, including straightforward ways to avoid them.

-- Dr. Seuss, How the Grinch Stole Christmas!, Random House, 1957

I first wrote about the Standard Template Library in 1995, when I concluded the final Item of More Effective C++ with a brief STL overview. I should have known better. Shortly thereafter, I began receiving mail asking when I'd write Effective STL.

I resisted the idea for several years. At first, I wasn't familiar enough with the STL to offer advice on it, but as time went on and my experience with it grew, this concern gave way to other reservations. There was never any question that the library represented a breakthrough in efficient and extensible design, but when it came to using the STL, there were practical problems I couldn't overlook. Porting all but the simplest STL programs was a challenge, not only because library implementations varied, but also because template support in the underlying compilers ranged from good to awful. STL tutorials were hard to come by, so learning "the STL way of programming" was difficult, and once that hurdle was overcome, finding comprehensible and accurate reference documentation was equally difficult. Perhaps most daunting, even the smallest STL usage error often led to a blizzard of compiler diagnostics, each thousands of characters long, most referring to classes, functions, or templates not mentioned in the offending source code, almost all incomprehensible. Though I had great admiration for the STL and for the people behind it, I felt uncomfortable recommending it to practicing programmers. I wasn't sure it was possible to use the STL effectively.

Then I began to notice something that took me by surprise. Despite the portability problems, despite the dismal documentation, despite the compiler diagnostics resembling transmission line noise, many of my consulting clients were using the STL anyway. Furthermore, they weren't just playing with it, they were using it in production code! That was a revelation. I knew that the STL featured an elegant design, but any library where programmers are willing to endure portability headaches, poor documentation, and incomprehensible error messages has a lot more going for it than just good design. For an increasingly large number of professional programmers, I realized, even a bad implementation of the STL was preferable to no implementation at all.

Furthermore, I knew that the situation regarding the STL would only get better. Libraries and compilers would grow more conformant with the Standard (they have), better documentation would become available (it has -- check out the bibliography beginning on page 297), and compiler diagnostics would improve (for the most part, we're still waiting, but Item 49 offers suggestions for how to cope in the meantime). I therefore decided to chip in and do my part for the burgeoning STL movement, and this book is the result: 50 specific ways to improve your use of C++'s Standard Template Library.

My original plan was to write the book in the second half of 1999, and with that thought in mind, I put together an outline. But then I paused and changed course. I suspended work on the book, and I developed an introductory training course on the STL, which I then taught several times to different groups of programmers. About a year later, I returned to the book, significantly revising the outline based on my experiences with the training course. In the same way that my Effective C++ has been successful by being grounded in the problems faced by real programmers, it's my hope that Effective STL similarly addresses the practical aspects of STL programming -- the aspects most important to professional developers.

I am always on the lookout for ways to improve my understanding of C++. If you have suggestions for new guidelines for STL programming or if you have comments on the guidelines in this book, please let me know. In addition, it is my continuing goal to make this book as accurate as possible, so for each error in this book that is reported to me -- be it technical, grammatical, typographical, or otherwise -- I will, in future printings, gladly add to the acknowledgments the name of the first person to bring that error to my attention. Send your suggested guidelines, your comments, and your criticisms to estl@aristeia.com.

I maintain a list of changes to this book since its first printing, including bug-fixes, clarifications, and technical updates. The list is available at the Effective STL Errata web site,http://www.aristeia.com/BookErrata/estl1e-errata.html.

If you'd like to be notified when I make changes to this book, I encourage you to join my mailing list. I use the list to make announcements likely to be of interest to people who follow my work on C++. For details, consult http://www.aristeia.com/MailingList/.

SCOTT DOUGLAS MEYERS
STAFFORD, OREGON
http://www.aristeia.com/
APRIL 2001