On Thu, May 1, 2008 at 9:22 AM, Burlen Loring <burlen.loring@xxxxxxxxxxx> wrote: > > To me it seems the compiler should not warn about code that follows the > ISO standard rules. However, it would be nice if the compiler warned about > the case that didn't follow the standard. I personally like to hide the > outer variables because it documents the fact that a particular loop does > not use that variable. Am I just being silly and I should use different > index variable name for all of my nested loops,even when there are no > dependencies? > I think that when the warning was introduced, it was very, very important, since it was noting a change in semantics. I'll agree that now it's becoming less useful, but it's still not quite been a decade since the standard, so there's quite plausibly still some legacy code around using the old behaviour. Perhaps a case could be made for removing it once -std=gnu++09 becomes the default, but even then, that kind of shadowing is generally a bad idea, and it's simple enough to remove -- and make more canonical, with an inner j -- by simply changing the relevant i's that I'm not convinced removing the warning is worthwhile. YMMV, ~ Scott
