[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: FileDescriptor proposal
From: |
Anthony Green |
Subject: |
Re: FileDescriptor proposal |
Date: |
23 Apr 2003 00:21:04 -0700 |
On Tue, 2003-04-22 at 22:29, Tom Tromey wrote:
> I think something similar to this can already be done with our
> existing approach.
Well, soon that should be possible. Unfortunately you can't just
implement FileDescriptor's public interface and pull your own
implementation in before libgcj. It still tries to pull both .o files
into the link and you end up with multiple definition errors. I don't
know why yet.
> It's unlikely that another
> method call here or there is going to be a big problem; I/O is slow.
Not necessarily (buffered IO or other in-core file representations).
> One open question is how your approach would work with java.nio.
I'm trying to figure out how the current scheme works.
java/io/FileDescriptor.java implements a package private constructor
that takes an int as the file descriptor. The comment says this works
because we access this method from C+++ where package private is
translated to public. However it looks like
gnu/java/nio/FileChannelImpl.java uses this method. How is this
possible?
AG
- Fwd: FileDescriptor proposal, Brian Jones, 2003/04/17
- RE: FileDescriptor proposal, Jeroen Frijters, 2003/04/17
- RE: FileDescriptor proposal, Anthony Green, 2003/04/17
- Re: FileDescriptor proposal, Tom Tromey, 2003/04/17
- Re: FileDescriptor proposal, Anthony Green, 2003/04/17
- Re: FileDescriptor proposal, Anthony Green, 2003/04/18
- Re: FileDescriptor proposal, Tom Tromey, 2003/04/23
- Re: FileDescriptor proposal,
Anthony Green <=
- Re: FileDescriptor proposal, Brian Jones, 2003/04/23
RE: FileDescriptor proposal, Jeroen Frijters, 2003/04/18
RE: FileDescriptor proposal, Jeroen Frijters, 2003/04/19