Re: floating point war casualty list

[dewar at gnat dot com (Robert Dewar)]

<<That's well and good.
The immediate reason I am here talking to you, however, is that people
are crying you broke my cray, you broke my pdp11, you broke my vax.
That is a regression in gcc-3.1.  The change that broke them
should be reverted.  According to your policy manual, they have
the right to insist that.
>>

Nobody has a right to insist on anything, unless they have a contract
that gives them that right. If you have a particular gcc port, especially
an obsolescent one, that you count on, you have three choices:

1. Use an older version that supports that target to your satisfaction

2. Fix the problem yourself

3. Make a contract with some third party to fix the problem

The policy manual, like the schedules you see, are merely internal goals for
a volunteer project. You can't insist on volunteers doingt anything. They
owe you nothing. In fact when you are dealing with volunteers, insisting
is probably counter productive. A more productive approach is to be positive
and let people know that you are using a particular port, and would find
it very helpful if it were fixed/continued etc. That way you may help to
encourage someone to do the work. Of course if you can volunteer to help,
that's even more constructive.