lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Yonik Seeley (Commented) (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (LUCENE-3610) Revamp spatial APIs that use primitives (or arrays of primitives) in their args/results so that they use strongly typed objects
Date Wed, 30 Nov 2011 01:27:40 GMT

    [ https://issues.apache.org/jira/browse/LUCENE-3610?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13159729#comment-13159729
] 

Yonik Seeley commented on LUCENE-3610:
--------------------------------------

bq. in either case: is there really any downside against having a simple object that wraps
the latlon pair so you always know which one is which?

We should always do lat first (and I believe we do).

bq. is the RAM/GC overhead of an object like i described really significant compared to a
double[2] ?

The object (compared to a double[]) could actually be better - no array bounds checks.  If
you're only talking about replacing double[2] w/ an object, I'm all for it.  Otherwise, it's
case by case - it depends if it's going to be used in an inner loop.
                
> Revamp spatial APIs that use primitives (or arrays of primitives) in their args/results
so that they use strongly typed objects
> -------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: LUCENE-3610
>                 URL: https://issues.apache.org/jira/browse/LUCENE-3610
>             Project: Lucene - Java
>          Issue Type: Improvement
>          Components: modules/spatial
>            Reporter: Hoss Man
>             Fix For: 4.0
>
>
> My "spatial awareness" is pretty meek, but LUCENE-3599 seems like a prime example of
the types of mistakes that are probably really easy to make with all of the Spatial related
APIs that deal with arrays (or sequences) of doubles where specific indexes of those arrays
(or sequences) have significant meaning: mainly latitude vs longitude.
> We should probably reconsider any method that takes in double[] or multiple doubles to
express latlon pairs and rewrite them to use the existing LatLng class -- or if people think
that class is too heavyweight, then add a new lightweight class to handle the strong typing
of a basic latlon point instead of just passing around a double[2] or two doubles called "x"
and "y" ...
> {code}
> public static final class SimpleLatLonPointInRadians {
>   public double latitude;
>   public double longitude;
> }
> {code}
> ...then all those various methods that expect lat+lon pairs in radians (like DistanceUtils.haversine,
DistanceUtils.normLat, DistanceUtils.normLng, DistanceUtils.pointOnBearing, DistanceUtils.latLonCorner,
etc...) can start having APIs that don't make your eyes bleed when you start trying to understand
what order the args go in.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

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


Mime
View raw message