commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gilles Sadowski <gil...@harfang.homelinux.org>
Subject Re: svn commit: r1166099 - /commons/proper/math/trunk/src/main/java/org/apache/commons/math/linear/LUDecompositionImpl.java
Date Wed, 07 Sep 2011 12:46:18 GMT
On Wed, Sep 07, 2011 at 01:16:03PM +0100, Mark Thomas wrote:
> On 07/09/2011 12:49, Gilles Sadowski wrote:
> > On Wed, Sep 07, 2011 at 11:39:37AM +0100, Mark Thomas wrote:
> >> On 07/09/2011 11:34, erans@apache.org wrote:
> >>> Author: erans
> >>> Date: Wed Sep  7 10:34:49 2011
> >>> New Revision: 1166099
> >>>
> >>> URL: http://svn.apache.org/viewvc?rev=1166099&view=rev
> >>> Log:
> >>> Added "final".
> >>> Moved declaration of "sum" where it is used.
> >>
> >> You might want to re-consider that. I suspect sum was declared outside
> >> of the inner loops to save allocating memory for a new double on every
> >> iteration of the loop. It may be worth moving the declaration outside
> >> the outer loop too.
> > 
> > Reducing scope makes for clearer code and can lead to _more_ optimized code.
> 
> Cleaner code, certainly.
> 
> As far as optimisation goes I'm not so sure with a primitive. On further
> research it looks like the performance should be identical. For an
> "interesting" discussion on what exactly is going on I found this:
> http://www.javalobby.org/java/forums/t16730.html?start=0

So:
 - Cleaner code: Yes
 - Faster code: Maybe
 - Higher memory usage: No

Do we agree that the change is fine?


Gilles

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
For additional commands, e-mail: dev-help@commons.apache.org


Mime
View raw message