This is the mail archive of the
mailing list for the Java project.
RE: FileDescriptor suggestion [Was: FYI: Patch: java nio file locking]
- From: "Boehm, Hans" <hans_boehm at hp dot com>
- To: "'Dalibor Topic'" <robilad at kaffe dot org>,Per Bothner <per at bothner dot com>
- Cc: Michael Koch <konqueror at gmx dot de>, java-patches at gcc dot gnu dot org
- Date: Mon, 26 Jan 2004 09:35:42 -0800
- Subject: RE: FileDescriptor suggestion [Was: FYI: Patch: java nio file locking]
The current garbage collector "knows" that memory is byte addressable,
and that addresses are 32- or 64-bits long and flat. (And it would be much easier
to teach it about 128-bit addresses than 16-bit ones.) If anyone wanted
to support Cray classic vector machines, or PDP 10s, or ..., a substantial
amount of GC work would be required.
> -----Original Message-----
> From: firstname.lastname@example.org
> [mailto:email@example.com]On Behalf Of Dalibor Topic
> Sent: Sunday, January 25, 2004 11:49 AM
> To: Per Bothner
> Cc: Michael Koch; firstname.lastname@example.org
> Subject: Re: FileDescriptor suggestion [Was: FYI: Patch: java nio file
> Hi Per,
> Per Bothner wrote:
> >> int in java is always 32 bit. Even on 64-bit systems.
> > > In contrary to C.
> > Out of curiosity, do you know of a target (that we're
> likely to support
> > for GCJ) where is C int is 64 bits? long is 64 bits on
> 64-bit systems,
> > but my impression is that the consensus is that int is 32 bits.
> Googling on newsgroups shows for example Crays as having 64
> buit ints. I
> don't know about gcj, but as there is one project that's
> porting kaffe
> to Cray, I doubt you'd want to be left behind ;)))
> The C standard doesn't maintain any consensus on int being 32
> bits, so
> implementations are likely to be 16, 32 and 64 (with the
> ocassional 18
> or 36 thrown in for good measure ;) bits.
> dalibor topic