[
https://issues.apache.org/jira/browse/MATH-348?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12841976#action_12841976
]
Ted Dunning commented on MATH-348:
----------------------------------
I agree that the usage should be consistent within a class. If instance
variables are used to compute the result, then the class should use only
instance variables. If getters are used instead, then they should be used
everywhere in the class.
I doubt very seriously that there is any measurable performance. JIT's are
very good a converting getters into in-line code.
That means that there are two possible courses of action:
a) avoid getters within a class
b) always use getters
Neither of these would be adversely impacted by sub-classing. In fact, I can
imagine that there would be a few cases where it would be very, very useful to
be able to over-ride the getter in order to force the class to do something
interesting. Those probably are corner cases that could be handled by using
the setters appropriately.
> Indirect access to instance variables
> -------------------------------------
>
> Key: MATH-348
> URL: https://issues.apache.org/jira/browse/MATH-348
> Project: Commons Math
> Issue Type: Improvement
> Affects Versions: 2.0
> Reporter: Gilles
> Priority: Trivial
> Original Estimate: 1h
> Remaining Estimate: 1h
>
> In all the methods (except the setters/getters) of
> BinomialDistributionImpl
> CauchyDistributionImpl
> ExponentialDistributionImpl
> FDistributionImpl
> GammaDistributionImpl
> HypergeometricDistributionImpl
> TDistributionImpl
> NormalDistributionImpl
> WeibullDistributionImpl
> ZipfDistributionImpl
> the instance variables are accessed through their respective getter.
> This is confusing (and possibly inefficient).
> What would be the expected behaviour of the getter if it were overriden?
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
