Re: What's Left for MingW 3.3

[Andrew Haley <aph at redhat dot com>]

Mohan Embar writes:
 > Hi Andrew,
 > 
 > > > - I don't think the logic for java.nio is correct on Linux either, even though
 > > >   the code happens to work
 > >
 > >The lack of a call to elements(), you mean?
 > 
 > I'm sorry. We're trying to talk about 3-4 different things at once and I'm not
 > helping much trying to keep things straight. From now on, I'll refer to specific
 > item numbers on my original list:
 > 
 > http://gcc.gnu.org/ml/java/2003-04/msg00166.html
 > 
 > The lack of the elements() call was item #2 on my original list, for which
 > I have a patch which I think you can commit with a peaceful conscience.

Yes.

 > What I meant by the logic in java.nio not being correct refers to item #1
 > on my original list.
 > 
 > ...which refers to this patch:
 > 
 > http://gcc.gnu.org/ml/java-patches/2003-q1/msg00670.html
 > 
 > Unfortunately, the patch consists of two unrelated pieces which in
 > my patch, I #ifdef-ed out. Michael later pointed out that for the
 > gnu/java/nio/natSelectorImpl.cc portion of this patch, I should
 > really implement a dummy _Jv_select() in win32.h instead of
 > #ifdef-ing out stuff in natSelectorImpl.cc. So I set out to do
 > this, but then I figured that since select() is also implemented on
 > Win32, I could just go ahead and try to do the real thing, which
 > opened up a can of worms and leads me to believe that the current
 > logic is faulty, even though the code works on Linux:

But it looks to me like this is quite deliberate:

  ServerSocket sock_object;
  int fd;

So there is a fd that belings to ServerSocket's impl and one that
belongs to ServerSocketChannelImpl.  I have no idea how all this is
supposed to work, so I suppose we need to get some input from the
author.

 > http://gcc.gnu.org/ml/java/2003-04/msg00185.html
 > http://gcc.gnu.org/ml/java/2003-04/msg00187.html

Okay, I'm back on course now.

I have a suggestion.

Please make a single patch that does the very *minimum* needed to
bring 3.3 to some sort of sanity on MingW.  This includes Michael's
suggestion about a dummy _Jv_select().  Then we can all look at it and
decide if it is OK, and we can test it, commit it, and move on.

Andrew.