On May 19, 2013, at 3:25 PM, Peter Collingbourne <[email protected]> wrote: > This function returns true if the ABI requires that we emit a > definition for a given function, given that we don't have a body > for it in the TU.
You can't emit a function when you don't have a body for it. What you're trying to do is force the emission of some kind of associated thunk because the ABI doesn't guarantee that one will be emitted in a canonical place. So what you should do is enhance the code that requires a reference to said thunk to force it into existence as necessary, instead of awkwardly shoving this into the deferred-declaration system. John. _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
