δΊ 2012/2/1 3:20, Ted Kremenek ει: > This all feels a bit piecemeal, which would be fine if we were fixing > specific issues, but the general sense I have right now is that how > destructors are currently represented in the CFG is possibly very different > from where we need to go once the infrastructure work in PR 11645 is done.
Agree. I thought of two approaches we can deal with temporary destructors. In one, we remain faithful to the AST. Whenever there is a CXXTemporaryObjectExpr, we create a dtor in the CFG for it. In the other one, we follow codegen, only creating dtors for those objects actually created (omit those elided objects). I incline to the latter one. _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
