Re: Run-time taint support proposal

Richard Lynch: > On Fri, December 15, 2006 4:31 pm, Wietse Venema wrote: > > Even if some taint check is to restrictive at some point in time, > > the programmer can always overcome it with an explicit action. > > Would that require some sort of bogus string non-manipulation, or do > you foresee a "untaint" function which does nothing but mark it > untainted, or are you thinking a new operator?... > > I think we've run out of ASCII symbols for operators, actually, so > don't pick that one. :-) Maybe we could use (tainted) and (untainted) typecasts as suggested elsewhere in this thread. For my proof-of-concept tests I implemented taint(), untaint() and is_tainted() extensions so that I could quickly set up a testbed. But if there is a better user interface then I have no problem. It's too early to say exactly when I will be able to share a first set of diffs with the community, but I'm sure it will be taken care of. Wietse