Re: RFT: The remainder of the InetAddress merge

[Gary Benson <gbenson at redhat dot com>]

Mohan Embar wrote:
> Mohan Embar wrote:
> > Gary Benson wrote:
> > > This patch is the second half of the InetAddress merge whose
> > > first half I committed earlier today.  It removes GCJ's custom
> > > InetAddress, and replaces the glue methods in VMInetAddress
> > > with real native code.
> > 
> > I am trying to build on Win32 with the patch you sent me off list
> > on 21 Sep and am getting this. Probably something simple like a
> > conflicting declaration but I won't be able to look into this more
> > until at least tonight: ...
> 
> I was able to work around the build errors by changing the offending
> lines (64 & 81) of natVMInetAddressWin32.cc to:
> 
>   JvSynchronize sync (&java::net::VMInetAddress::class$);
> 
> (added address-of operator (&) - don't why the POSIX stuff builds
> okay without it).

Hmmm, no idea :)  But I added it to the Posix methods too...

> The build worked and passed the attached test, which is more testing
> than this class has probably ever offically gotten on Win32. Your
> patch gets my thumbs-up for the Win32 portion.

Thank you.  I'll post an updated patch shortly.  With luck we can get
this into 4.2.

> I haven't tested the security portion of this.  Let me know if you
> think that's necessary.

It shouldn't be.  The security checks are all in Java now, shared
across all platforms, so testing on any one of them should be
sufficient.

Thanks again,
Gary