Displaying 3 results from an estimated 3 matches for "pr24078".
2015 Jul 16
10
[LLVMdev] [RFC] Defining Infinite Loops
...'finite', but bikeshedding is welcome.
With such an attribute in place, we would be able to clarify our overall position on infinite loops, be in a stronger position to infer more specific function properties (like halting), and can put in place generally-correct fixes to outstanding bugs (PR24078, for example). I know there are some Clang users who want it to optimize while honoring infinite loops, and I think adding this attribute helps them as well (assuming we'd provide some non-default option to prevent Clang from adding it). Thoughts?
Thanks again,
Hal
--
Hal Finkel
Assistant Co...
2015 Jul 16
2
[LLVMdev] [RFC] Defining Infinite Loops
...eshedding is welcome.
>
> With such an attribute in place, we would be able to clarify our
> overall position on infinite loops, be in a stronger position to
> infer more specific function properties (like halting), and can put
> in place generally-correct fixes to outstanding bugs (PR24078, for
> example). I know there are some Clang users who want it to optimize
> while honoring infinite loops, and I think adding this attribute
> helps them as well (assuming we'd provide some non-default option to
> prevent Clang from adding it). Thoughts?
>
> Thanks again,
&g...
2015 Jul 16
2
[LLVMdev] [RFC] Defining Infinite Loops
...; >
> > With such an attribute in place, we would be able to clarify our
> > overall position on infinite loops, be in a stronger position to
> > infer more specific function properties (like halting), and can put
> > in place generally-correct fixes to outstanding bugs (PR24078, for
> > example). I know there are some Clang users who want it to optimize
> > while honoring infinite loops, and I think adding this attribute
> > helps them as well (assuming we'd provide some non-default option
> > to
> > prevent Clang from adding it). Thought...