On Feb 13, 2009, at 9:50 AM, Daniel Dunbar wrote: > However, since this is a much more significant functionality change, > that strikes me > as another pretty bad idea.
I'd say, get over it. gcc handles this by having a bit, that is 0 in the development trees, and 1 in release branches, and it sets the default for all sorts of fancy checking to this bit. The advantage is faster compile times, which is desirable, easy configuration, most of the time, fast compiles on release, and slow, but more checking, during development. This essentially mirrors using NDEBUG. In gcc, there is more control, and one can configure with more or less checking, and there are many bits, each controlling a different aspect of checking. Very expensive things, never default to on. Very cheap things, rarely turn off. If we only had 1 driver (I don't see why we should have more than 1), trivially, we could not do this in the style that is done today. _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
