RE: New IO Native Provider Interface

[Jeroen Frijters]

Hi,

Great job, this is almost exactly like I envisioned it. A few minor
suggestions:

- FileOutputStream should not request read access, so the modes should
be "w" and "a"
- write(byte[] buf, long offset, long length) should not use longs for
offset and length
- nativeReadByte shouldn't return a long
- I don't think available() should return negative numbers

I'm still in doubt as to whether there should be a VMFileDescriptor. For
now, I actually reimplemented FileDescriptor and because the public
interface is fairly small, I feel this works well and it saves the
overhead of another layer of indirection.

Regards,
Jeroen

>  
>  
> -----Original Message-----
> From: Aaron M. Renn [mailto:address@hidden 
> Sent: Tuesday, March 04, 2003 23:44
> To: address@hidden
> 
> I modified the java.io package to do native I/O (except for 
> the File class)
> along the lines of what I posted last week.  Please take a 
> look and let
> me know ASAP if anything is broken.  I'm especially keen to 
> hear from the
> gcj people on how close this comes to allowing us to merge 
> these classes.
> This also corrects the problems recently discussed of finalization in
> classes like FileInputStream since right now only FileDescriptor deals
> with native file descriptors and native methods.
> 
> -- 
> Aaron M. Renn (address@hidden) 
> http://www.urbanophile.com/arenn/
> 
> 
> 
> _______________________________________________
> Classpath mailing list
> address@hidden
> http://mail.gnu.org/mailman/listinfo/classpath
>