From: Warren Young Date: March 8 2006 8:38pm Subject: Re: Transaction class design in Wishlist List-Archive: http://lists.mysql.com/plusplus/5537 Message-Id: <440F40E0.6060203@etr-usa.com> MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Drew Vogel wrote: > > "Too bad this isn't > C++, because then I could just subclass std::list. I can't say I'm wild about that idea. First, that implies multiqueries, which return results in a harder to handle way than regular queries. (See examples/multiquery.cpp.) Second, there's the is-a issue: a transaction set is not a special kind of query. It's more of a "flag" temporarily set on a per-connection basis, and later cleared. I could see someone wanting to implement transactions by extending Connection for this reason. But I think the separate Transaction class is cleaner. > I'm mentioning this, not to criticize your efforts, but because I will > be implementing this in the semi-near future. Would anyone on the list > be interested in it once it's done? Go ahead and post it to the list. It's more productive to argue over working code than design points. But before you start writing code, why don't you check out the v2.1 branch and see what you think of the new Transaction class? If you don't want to play with svn, 2.1 alpha 4 should be out within a few days, and it will include this.