A completely different approach to EH runtime

[Mike Stump]

> Date: Wed, 21 Feb 2001 13:38:17 -0800
> From: "H . J . Lu" <hjl@valinux.com>
> Cc: Zack Weinberg <zackw@Stanford.EDU>, gcc@gcc.gnu.org

> I don't think it will be a problem for you. But do you think the Red
> Hat tech support people like to hear "I install a new gcc and my
> machine won't work anymore"?

Why is it impossible to run a mini testsuite to catch major problems
at install time, and only install, if the new one passes the
testsuite?  Every program you worry about breaking could then have a
testcase in the testsuite, and install will only put it into lib
_after_ the testsuite runs with the new lib it is about to install.

Why can this approach not catch most of the common problems you can
forsee?