I know. But once I add the ability to replace a mutex there's no way to detect how it's being used. Though as its in core I guess the user should simply be trusted to do the right thing.
Sent from my iPhone On Jan 9, 2012, at 6:57 AM, "Steven Schveighoffer" <[email protected]> wrote: > On Mon, 09 Jan 2012 09:41:40 -0500, Sean Kelly <[email protected]> wrote: > >> Which goes back to my argument that we should ship both checked and >> unchecked (release and not) builds and the compiler should link the proper >> one. > > Or you make it a release-mode check. I can't imagine this being an > inner-loop operation. > > But my point was, it's already unsafe. We've not had horrible issues with it > in its current state. Is it worse to allow replacing the monitor? If you > take the right steps (i.e. do it first line of the ctor), it's not going to > result in a race condition. > > -Steve
