[jira] [Commented] (PDFBOX-1979) TypeTestingHelper is non-deterministic

["Tilman Hausherr (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/PDFBOX-1979?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14177094#comment-14177094 ] 

Tilman Hausherr commented on PDFBOX-1979:
-----------------------------------------

While your patch makes it deterministic, the test cases are now too narrow, i.e. within an interval. What would be better would be pseudorandom that are deterministic (e.g. start with a fixed seed), AND "surprising" random tests (i.e. the current code) as a bonus, like done in PDFBOX-1977 (there's a long discussion there and the consensus was that we need deterministic and non-deterministic) and the two follow-up issues.

> TypeTestingHelper is non-deterministic
> --------------------------------------
>
>                 Key: PDFBOX-1979
>                 URL: https://issues.apache.org/jira/browse/PDFBOX-1979
>             Project: PDFBox
>          Issue Type: Bug
>          Components: XmpBox
>    Affects Versions: 1.8.7, 2.0.0
>            Reporter: John Hewson
>            Assignee: Guillaume Bailleul
>            Priority: Minor
>             Fix For: 2.0.0
>
>         Attachments: nd_test.patch
>
>
> TypeTestingHelper generates random calendar data and random UUIDs for testing, which means that it is non-deterministic.
> As discussed in PDFBOX-1977, we should alter this test to make sure that it has deterministic (regression test) functionality as well as the existing non-deterministic (fuzz test) functionality.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)