Strong disagree. Korn's response is at least funny. The rest of the responses are petulant, somewhat ill-informed, and orthodoxical. The argument over whether C does or doesn't have a runtime is particularly tedious. You get the impression that these people would all be very, very upset at the Detours paper, except that Detours works so well they'd know a priori they'd lose the argumet.

The original commenter --- despite his very fragile code --- is right: for what he wants to do, he needs to create temporary stacks. What he doesn't have is a mechanism to allocate and swap in and out of those stacks (he's hoping that switching stacks is a 2-liner that just relies on knowing the inline asm incantation for changing ESP).

Regardless, it is a uniquely and flagrantly bad way to educate (both the questioner and the world who's reading the thread) to start from the premise that "you're not ready to build the thing you want to build" --- at least when "the thing" we're talking about is systems code. 90% of C programmers are not "ready" to build production C code that will get deployed in hostile environments, and as a result, in every web project we ever do, we have to embark on a tedious inventory of all the 3rd party stuff our clients use to find the horrible C ZIP library they accidentally included. No C coder is blameless.

Unless I missed a whole leg of this thread (the part where they explained temp stacks and stack switching to this guy, who, in challenging the notion that C has a real runtime or the notion that the OS should [heh] be controlling ESP, is demonstrating more on-the-ground systems programming knowhow than several of the commenters), I think this is a remarkably douchey example of a programming help thread.