[
https://issues.apache.org/jira/browse/NUMBERS-38?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16036266#comment-16036266
]
Gilles commented on NUMBERS-38:
-------------------------------
bq. I think all that you copied from commons math is correct just need better
refactoring of code
I'm now confused as to what the problem is.
Either we talk of *bugs* or we talk about design. The refactoring was based on
the simple premise that each building block (aka a function that has a name and
is the implementation of a reference algorithm) is defined in its own class.
If the output of Gamma is buggy, please file a JIRA report and then create a
corresponding unit test.
bq. I Gamma we are generating values with basic logic if n <= 20 else we
calculate it with LanczosApproximation, all we can do is to put full logic in
LanczosApproximation.java and just call it, at this point Test-cases on
LanczosApproximation doesn't make sense.
I don't follow. Why couldn't you generate a list of reference output values,
from any of the codes you've found, that checks the correctness of
{{LanczosApproximation}} separately from the classes that make use of that
output?
> No unit tests for "LanczosApproximation" class
> ----------------------------------------------
>
> Key: NUMBERS-38
> URL: https://issues.apache.org/jira/browse/NUMBERS-38
> Project: Commons Numbers
> Issue Type: Test
> Reporter: Gilles
> Labels: unit-test
> Fix For: 1.0
>
>
> The computation of the {{LanczosApproximation}} (package
> {{o.a.c.numbers.gamma}} in module {{commons-numbers-gamma}}) function is not
> checked by unit tests.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
