Thanks for working on this so diligently guys!

On Tue, Jun 4, 2013 at 9:06 AM, Travis L Pinney <[email protected]>wrote:

> Thanks for the suggestions. I will submit another patch.
>
> Cheers,
> Travis
>
>
>
> On Tue, Jun 4, 2013 at 5:11 AM, Martin Desruisseaux <
> [email protected]> wrote:
>
> > Le 04/06/13 01:12, Travis L Pinney a écrit :
> >
> >> Currently I have it set up to MMAP the files. I am still using Apache
> >> EndianUtils for reading Little Endian values. I am not sure of the
> >> advantage of calling foo.order(ByteOrder.LITTLE_**ENDIAN) instead of
> >> using
> >> the EndianUtils, (...snip...)
> >>
> >
> > I noticed that you switched to
> ByteBuffer.order(ByteOrder.**LITTLE_ENDIAN),
> > thanks. For the record, the advantages are:
> >
> >  * No need for external libraries, since ByteBuffer is bundled in JDK
> 1.4+
> >  * No need for additional API, since the same set of existing methods
> >    work in both cases
> >  * Potentially order of magnitude faster, since ByteBuffer.get*() with
> >    Little Endian order maps directly to a single CPU instruction on
> >    Intel processors, while I guess (but did not verified) that
> >    EndianUtils performs arithmetic operations for achieving equivalent
> >    result.
> >
> >
> >     Martin
> >
> >
>

Reply via email to