Axel Dörfler <axeld@xxxxxxxxxxxxxxxx> wrote: > "Jonas Sundström" <jonas@xxxxxxxxxxx> wrote: ... > > I would like to consider myself done, if there's no > > outstanding issue. Feel free to make changes, though. > > Unfortunately, you just removed my remark about the current > LockBuffer(0) strategy; I'm happy to hear any arguments that > would make the current solution worthwhile, personally, I think > it only covers a very unlikely special case, and is inefficient > in the general case -- for a generic API, the usual aim is the > opposite, though. I read it and I agree of course. I'm not saying I consider the method optimal in its present form. I just got tired. I'll change it back to initially use an on-stack buffer, 1024 in size. /Jonas.