Re: Patch: New implementation of -Wstrict-aliasing

[Dirk Mueller <dmueller at suse dot de>]

On Thursday, 1. February 2007 18:25, Benjamin Kosnik wrote:

>  > > I much like this scheme where higher levels give more accurate
>  > > diagnostic, and still level room for improvement did we need to add
>  > > more levels (I hope that need would not be frequent, still).
> I've got to say it: me too. I really like the idea of verbosity control.

What does it mean? So far we're using levels for "the higher the number, the 
more warnings will be triggered, and the more likely it is that it is a false 
positive". 

If you're going to reverse that, then its going to be a hell of a lot of work 
for a user to maintain the "level": if we add new levels, he has to write a 
configure check that, depending on the compiler version, changes the warning 
level. That doesn't make sense.

> -Wfortify

See my other patch, it is mostly finished (need to write some testcase and 
prepare a proper submission). 

> -Wpedantic=1-10

hmm. What if I like the pedantic warnings of level 1 and 3, but hate level 2?


Dirk