On Friday 25 October 2002 9:28 am, Lars Gullik Bjønnes wrote: > Angus Leeming <[EMAIL PROTECTED]> writes: > | On Friday 25 October 2002 8:59 am, Lars Gullik Bjønnes wrote: > | > Angus Leeming <[EMAIL PROTECTED]> writes: > | > | John, > | > | > | > | attached is some rough and ready, proof of principle > | > | code that appears to do the job. > | > | > | > | To generate an autosave type > | > | buffer-auto-save > | > | in the minibuffer. You should get "Autosaving Buffer" > | > | printed to the console. If you have the Child Processes > | > | dialog open, you should also get something flashing up > | > | there, indicating that the process is being stored by > | > | the ForkedController as we desire. > | > | > | > | I certainly can't generate zombies, but maybe you'll > | > | have a go too? > | > > | > Ok, Good work. > | > > | > ...but I am not sure that I agree with the interface. > | > Especially the ForkedBase(?). > | > > | > I was imagining some interface like: > | > > | > processController.call(boost::bind(autoSave, > | > &buffer), Proc::background); > | > > | > > | > which would fork, and run the function object in the newly > | > created process. > | > | I tried that. The problem is that we'd need > | > | template<typename Func1, typename Arg1> > | processController::call(Func1 & func, Arg1 const & arg1); > | > | template<typename Func2, typename Arg1, typename Arg2> > | processController::call(Func1 & func, Arg1 const & arg1, > | Arg2 const & arg2); > | > | etc. > > No you wouldn't. All that would be hidden with the > boost::function functor. > > processController::call(boost::function0<void> func); > > should be enough. > > and then boost::bind would be used to reduce whatever function > to that signature. > > boost::function<void> func = boost::bind(autoSave, arg, arg > ,arg); > > func(); // call without args, they are already bound.
Oh, good man! Thank you. I'll put this in my secret stash of things for later. > | Feel free to prove me wrong here, but my proposed solution > | is little more than rearranging code that we know works. We > | are in bug-fixing mode, remember. > > as such it is quite ok, but there is a bit more coupling now, > isn't it? Probably. But I'll get it working as it is because it will apply cleanly to 1.2.X also. Angus
