Re: Support cross-testing

[Roland McGrath <roland at hack dot frob dot com>]

I fully support the goal and I'm grateful that you are working on it now.
I don't have time right now (nor all that soon) to give a thorough review
of the details.  This is a large change touching a lot of subtle stuff.  It
would certainly be easier if you can find a way to do it piecemeal.  (The
nm example is obvious and trivial, so there's absolutely no reason not to
just do that first.  Perhaps there is some other similarly low-hanging
fruit, though I have no such ideas off hand.)

>From a quick glance, the repetitive nature of many of the changes suggests
that there is probably some substantial amount of refactoring of the way we
have things organized today that might well consolidate some of the work
being done (and changed here) into fewer places that the details need to be
maintained.  But that's a very vague intuition and it might be significant
work just to puzzle it out.

In the pro forma trivia department, I'm not at all sure about the propriety
of crediting someone in ChangeLog entries using an email address he no
longer has (especially when it's with a date years after he did the
original work, and when the work you are submitting differs much at all
from what he did originally and he was not involved in writing or reviewing
the current version).  I'd suggest you just ask Jim whether he prefers to
be credited with the obsolete address, credited with a different address,
or not formally credited at all.  ("Informal credit" would involve just
mentioning his name, perhaps with no address at all, in the log text but
not in the header.)  I'm fairly sure Jim is not interested in contributing
to the review or reworking of the current version of the change (having
seen him just the other day).


Thanks,
Roland