On Thu, Apr 04, 2002 at 03:48:22PM +0200, Lars Gullik Bjønnes wrote: > | [Which is always a pain in the ass to maintain if you have a dozen members > | or so...] > > ... and if there are a lot of class variables then something is wrong > with the design... (I think I even heard you say that) > > And we want to be typesafe!
Hm, yes. But what as 'typesafety' with the existence of a handcrafted operator=() to do? > | What the heck is wrong with the auto-generated version of operator=(). > | Less code means less maintance. > > auto generated versions of operator= are never typesafe. ? Do you mean something like 'threadsafe' or 'commit-or-roll-back'? 'typesafe' makes absolutely no sense to me in this context. > | And don't tell me LyX would do anything sensible > | if constructors start throwing. > > Sure we could. Gracefull shutdown. emergency save. delete some > structures to get more memory. We get the shutdown and emergency save already with catching a signal or two or just by providing our own allocator. I would not ever consider implementing 'deleting some structures' for _LyX_. This is almost impossible to get right in the first place and saves what? Two keystrokes to move up in the shell history an <Return> after a clean shutdown? You are kidding, arne't you? > | I see no benefit in adding hunderds of lines of code just for obscure > | theories.. > > so so obscure... and not theories either... Every now and then somebody adds a member to the class but not to swap. No compiler warning possible. Been there, done that. Removed swap again in most cases. Called it a 'obscure theory'. As long as there is no standard support for std::swap (like auto-generating it as member-wise swapi e.g.), it is an endless source of trouble. And as I understand the gurus, there never will be. Andre' -- Those who desire to give up Freedom in order to gain Security, will not have, nor do they deserve, either one. (T. Jefferson)
