Re: Small update to reversed_comparison_code

[Mark Mitchell <mark at codesourcery dot com>]

>>>>> "Richard" == Richard Kenner <kenner@vlsi1.ultra.nyu.edu> writes:

    Richard>     There may be situations where everyone agrees that
    Richard> (say) the SPARC back-end is broken, and that it is better
    Richard> to break the SPARC bootstrap for a while in favor of some
    Richard> patch that does something good.

    Richard> That's basically what I'm saying.

Right -- but the key point is *when* we should do this.  There are
certainly cases where the breakage is OK, and certain some where it is
not.  I think most of us believe that breakage should be very rare.

    Richard> It's impossible to fix a bootstrap problem using a
    Richard> cross-compiler unless somebody who has the target machine
    Richard> has identified what was miscompiled, for example.  But
    Richard> it's not just lack of a physical machine, but the issue
    Richard> of whether the submitted can be expected to have
    Richard> familiarity with some of the more obscure ports.

Yes, a bootstrap failure can be hard to track down.  And yes, you'll
need help sometimes.  And yes, obscure platforms matter less.

So, if your point is that things are a matter of degree, that's true.

But, over the last several weeks we've had broken 3.0 branch
bootstraps on Solaris, and broken mainline bootstraps on x86.  Those
*are* major platforms, and both have remained broken for *weeks*.

I think most people would agree that's unacceptable.

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com