Re: don't assume pointer cast to unsigned long is a valid initializer

[Mike Stump <mrs at apple dot com>]

On Monday, March 3, 2003, at 12:10 PM, Alexandre Oliva wrote:
> On Mar  3, 2003, Geoff Keating <geoffk at geoffk dot org> wrote:
> > You're saying it doesn't work.  I said it should work.
>
> I've already explained that there are ports in which it can't possibly
> work, and on most ports your test is redundant with another test we
> already have.  You're trying to force the world to accept your notion
> that every post must have a relocation that is as wide as a long.

My take on this is, this isn't a testcase for determining ANSI C 
correctness.  I think we can agree on this.  So, what is it then?  It 
would be a testcase that tests a gcc feature.  Tests that test gcc 
specific features have been put into the testsuite and are reasonable.  
One cannot argue that just because this doesn't have ANSI C mandated 
behavior it cannot be in the testsuite.  The next interesting question 
is, is the feature intended to be portable across all machines?  If 
yes, the testcase is ok as it was.  If it is intended to be portable to 
most, it should be ok as is.  If it is portable to a really limited 
subset, then the code should be changed to

#ifdef SUBSET
...
#endif

or some other such spelling.  Now, if one person wants gcc to have a 
feature and another person doesn't want gcc to have a feature, they 
should discuss it, and failing agreement, they should ask for input 
from the rest of the community, and if they still cannot accept 
agreement, they can appeal to the SC.  In the mean time, status quo 
should probably prevail if the disagreement is with two write privs 
people, and in favor of the write-privs person otherwise.

By having a process that is fair and reasonable for the thorny 
situations, we don't have to feel completely frustrated in these 
situation, but rather, the decision was X, and that is just how it is.  
If people feel wronged by the process, let's talk about the process 
apart from the technical issue at hand and see if we can agree on what 
it should be.  Above I give my take on what I think the status quo 
process is (or should be).

Personally I am not offended by the notion of forcing every port to 
have a relocation as wide as a long.  I don't think this is completely 
unreasonable.  If this is really the issue, this is what we should be 
discussing.