This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH] Document arithmetic overflow semantics
On Thu, Feb 13, 2003 at 10:41:35AM -0700, Roger Sayle wrote:
> How? Any new optimization that we introduce that takes advantage
> of the undefinedness will both produce different results from the
> current versions of GCC,
Of this I do not care.
> ... and will break the Java front-end.
The Java front end is already broken because of the overflow
assumptions we _already_ make. See extract_muldiv.
This is why we should have proper annotations on the tree nodes
so that Java can get the results that they need.