This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Patch: New implementation of -Wstrict-aliasing
On Sun, 4 Feb 2007, Silvius Rus wrote:
| On 2/3/07, Gerald Pfeifer <gerald@pfeifer.com> wrote:
| > On Thu, 1 Feb 2007, Gabriel Dos Reis wrote:
| > > No, we're using the number to mean "the higher the level, the more
| > > accurate it is; e.g. less false positive". Essentially because higher
| > > levels demand more computations.
| >
| > This is not consistent with -Wstrict-aliasing and -Wstrict-aliasing=2
| > as currently defined.
|
|
| Hello Gaby and Gerald,
|
| My email on January 30 proposed warning levels in the order
| recommended by Gerald. My intention was to be consistent with the
| current direction (level 2 = more verbose than 1). However, after
| Gaby's comments, I believe that the order should be (higher level =
| more effort). This will allow us to add more levels when/if more
| advanced analysis becomes available.
Hi Silvius,
After Gerald's mail, I went back and reviewed the setup for
-Wstrict-aliasing. My conclusion is that it is not be a good model
for your proposed warning. The new scheme allows us more flexibility.
-- Gaby