Dr. Dobb's Journal - December 2007 - (Page 56)

d12gran_p5ma 10/12/07 10:29 AM Page 56 Core Technology TRANSACTIONAL PROGRAMMING Container list list vector vector set set map map Operation insert delete insert delete insert insert insert delete insert delete insert delete insert delete std:: (ms) 750 297 3469 985 16 3688 1172 375 5765 1062 2203 375 9517 1375 atomic:: (ms) 937 313 3718 937 375 4516 1484 390 5890 1016 2452 390 9813 1391 atomic/std 1.25 1.05 1.07 0.95 23.4 1.22 1.26 1.04 1.02 0.96 1.11 1.04 1.03 1.01 Table 1: Measuring the performance of transactions. transaction stack. The extra memory used is proportional to the size of the transaction, not the size of the data. Provided that transactions are short, little extra memory is required. Object deletion is deferred until the end of the transaction. Table 1 shows some performance comparisons of atomic containers compared to their std equivalents, in nonthrowing circumstances. The program used to generate these numbers is available online (see “Resource Center,” page 5) and at calumgrant.net/atomic. These figures measure the same operation performed repeatedly on different types of containers. As you can see from these results, atomic containers do add some overhead, which is to be expected. The only container for which the overhead dominates the performance is for vector , which is so efficient that the operations on the transaction stack are slower than the operations on the vector itself. Conclusion The problem with C++ exceptions is that they can occur at many points in a program, and if even a single exception is overlooked, the program is potentially unsafe. Database transactions don’t suffer from this problem because atomic behavior is guaranteed by the transaction. By copying the database approach, C++ can also implement transactions. It provides guarantees of atomic behavior, helps in complex error-recovery situations, and allows smaller functions to be composed into larger transactions. Whichever exception strategy is used, it is an important part of the design and needs care, attention, and planning. References [1] Abrahams, D. “Exception-Safety in Generic Components,” Generic Programming: Proceedings of a Dagstuhl Seminar, M. Jazayeri, R. Loos, and D. Musser, eds. (Springer Verlag, 1999). [2] Alexandrescu, A., P. Marginean. “Change the Way You Write ExceptionSafe Code—Forever.” (www.cuj.com/ documents/s=8000/cujcexp1812alexandr/ alexandr.htm). DDJ 56 Dr. Dobb’s Journal l www.ddj.com l December 2007 http://www.programmingresearch.com http://www.cuj.com/documents/s=8000/cujcexp1812alexandr/alexandr.htm http://www.programmingresearch.com http://www.cuj.com/documents/s=8000/cujcexp1812alexandr/alexandr.htm http://www.cuj.com/documents/s=8000/cujcexp1812alexandr/alexandr.htm http://www.ddj.com

Table of Contents Feed for the Digital Edition of Dr. Dobb's Journal - December 2007

Dr. Dobb's Journal - December 2007 Contents Hmmmm Alia Vox Developer Diaries Developer’s Notebook Computer Books: Reading Between the Lines Conversations Query Anything with SQLite XQuery Web Maps with the Google Map API OpenALM and Its Manifesto Transactional Programming Effective Concurrency The Agile Edge Swaine’s Flames

Dr. Dobb's Journal - December 2007

For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.