OK given that it is only better error detection then a warning is not needed.   I mistakenly assumed policy might also be applicable to other compiler settings like speed.


On 6/29/07, Nikodemus Siivola <nikodemus@random-state.net> wrote:
Liam Healy wrote:

> Would it be possible to have a warning or note issued when a declaration is
> overridden due to the policy?  Many a head-scratching debugging has
> resulted
> when a global setting overrides a local one.  My experience with emacs
> programming for example is that I set a variable I think does what I want,
> and there is absolutely no effect because of some global setting I didn't
> realize took precedence.

I'm not sure I follow you.

Given that the _only_ thing RESTRICT-COMPILER-POLICY does is override local
optimization qualities (adjusting them upwards), giving a warning for that
seems rather redundant. Also: only change in semantics that R-C-P can cause
is better error-detection.

Can you give an example of the kind of confusion you envision, and how
a warning would help you solve it?


   -- Nikodemus