Allen Briggs wrote:

> I'm against flipping it when we enter RC.  I'm not against flipping
> it at the branch point.

These minorities are of less interest to me; what I care about is that
the majority of the people who use -current (and whether it's just
-current or the RC tags as well) will, by default, run with DIAGNOSTIC.

> This case is a bit of an exception as I noted before.  I think enough
> people will disable it in custom kernels that it will get a reasonable
> amount of testing during the regular development cycle in -current.
> This is not true of many options, but I think it's true of this one.
>
> You should ensure that we are not introducing any API changes when
> switching to/from DIAGNOSTIC (as Chuck noted).

Absolutely; do we know (except for malloc) which parts specifically are
affected? I agree this should be addressed regardless of what happens
with DIAGNOSTIC.

-e.

-- 
Elad Efrat