On Jul 22, 2012, at 11:33 AM, Jonathan Sauer wrote: > turns out the fix is even simpler: As the whole "POD for the purpose of > layout"-business is > already taken care of by the layout builder (albeit without considering > bitfields ATM), there > is no need to duplicate the logic in the new getTypeInfo-method: It simply > fetches the record > layout and returns its data size. > > I am wondering if the size of a record as returned by getTypeInfo could > always be the data > size, but I'm not knowledgable enough to assess this.
No, most clients care about the size of a complete object. getDataSize() doesn't just drop tail padding; it also drops virtual base subobjects. Did you do any bootstrap performance investigation? Because like I said, I had a patch like this before, and it had to be reverted due to bootstrap performance problems. John. _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
