Successful Build and Install

[Joe Buck]

I wrote:

>   > It seems that we would only want:
>   > * reports of failures on systems we believe to work
>   > * reports of successes *or* failures on systems we haven't tested.

Jeff writes:
> This is probably the right general direction for minor releases.
> 
> However at each major release it's probably useful to get these
> reports for all systems -- including those we regularly work on.

Just the same, I'd like to cut down on email volume.  We can expect
the egcs user community to increase greatly with time.  Do you want
every single egcs user on Linux to send a mail to everyone on the list,
saying essentially the same thing?  You'll start to drive away developers.
People have other things they want to do with their lives than read egcs,
and if they try using filters they'll miss important messages (delete
all test results?  what if some expose problems?).

Another alternative is to publish a summary of test results on platforms
that we have tried.  We can then ask people to send in a report if their
results differ *and* they've checked a FAQ to make sure they haven't made
a common mistake.  If we have no report for their system, we would use
whatever users send in.

> It's been my experience that as a developer I know how to build
> things dead cold.  It's amazing what kinds of problems are exposed
> by folks that don't build compilers every day...

Exactly.  That's why we need reports from people who get different
results.  But we don't need reports from people who get the same results.