Submitted a configurable and slightly more conservative version as r193410. Let me know if it needs further tweaking, ideally with examples.
On Thu, Oct 24, 2013 at 8:30 PM, Daniel Jasper <[email protected]> wrote: > Yes, I imagin an integer-value. But finding the "right" value is hard. In > the patch's current form, it uses the same as for function definitions. And > in function definitions, it is definitely worse, as they more often look > like a "wall of text" then. I would start with half the penalty and see how > it goes. > > > On Thu, Oct 24, 2013 at 8:26 PM, Chandler Carruth <[email protected]>wrote: > >> On Thu, Oct 24, 2013 at 11:22 AM, Daniel Jasper <[email protected]>wrote: >> >>> Yes, his change does that. All that I am asking is to not hard-code this >>> but instead introduce a style-dependent value. I can pick this up >>> tomorrow.. Should be a very straight-forward patch. >>> >> >> I'm also wondering whether we really want to be able to tune this more >> precisely. IE, it may have overcompensated in its current form.... >> > >
_______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
