Re: Small update to reversed_comparison_code

[Mark Mitchell <mark at codesourcery dot com>]

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

    Richard> I wouldn't consider a miscompare during a bootstrap as
    Richard> "not being able to build anything".

Technically, you are correct of course.

However, if a developer checks out, and cannot do `make bootstrap'
then it is difficult to test on that platform.

    Richard>     The right solution is to hold on to the patch, and
    Richard> Yes, but is that enough?  Look how many people have
    Richard> contributed to tracking down this problem in recent days?
    Richard> Would that ever have happened if the patch had
    Richard> immediately been removed?  Who would bother?

Perhaps Richard himself, since the hard work he put it into his patch
would have been for nought!

    Richard> I certainly agree that the tree should work on major
    Richard> platforms most of the time.  My argument is that trying
    Richard> to attain a standard where it works on most (not just
    Richard> major) platforms *all* of the time will have the effect
    Richard> of hindering development and keeping latent bugs hidden.

Again, a question of degree.

    Richard> I agree that the case of an outright bug in somebody's
    Richard> change is a different situation, but that's not been the
    Richard> most common problem.

Actually, it has.  There are many cases where a check-in ignores
CC0-machines, or machines that don't use DWARF, or C++, or some other
similar such.

I don't see continued debate here as fruitful -- clearly you do not
agree with me.  Going forward, we shall simply have to see what the
FSF, the SC, and the community decide to do.

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