On Tue, Oct 22, 2013 at 10:17 AM, Diego Novillo <[email protected]> wrote:
> On Tue, Oct 22, 2013 at 1:13 PM, Chandler Carruth <[email protected]> > wrote: > > > I'm surprised -- I expected this to essentially only be used with > > sample-based profiles? Bob seemed to have a different architecture in > mind > > for other profile sources? > > For now, sure. But there are other profile sources that are not based > on sampling. Instrumentation being one. Value tracking, being > another. AFAIR, Bob is looking at instrumentation-based approaches. > > > If this is going to be something generic, we should have some input from > Bob > > as well at least. And then the flag name it seems should be "-fprofile" > or > > "-fprofile-file" or "-fprofile-input".... > > Sure. No qualification seems the best option here. I wonder if the better way to go is to start with a very narrow flag name, and generalize it when the infrastructure for more generalized profile reading arrives? Supporting a highly specific flag for a long time doesn't seem like a high cost.
_______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
