[Octave-bug-tracker] [bug #65406] pkg test statistics fails include testing shadowed functions with core equivalents?

[Nicholas Jankowski]

URL:
  <https://savannah.gnu.org/bugs/?65406>

                 Summary: pkg test statistics fails include testing shadowed
functions with core equivalents?
                   Group: GNU Octave
               Submitter: nrjank
               Submitted: Sat 02 Mar 2024 11:15:29 PM EST
                Category: Test Suite
                Severity: 3 - Normal
                Priority: 5 - Normal
              Item Group: Unexpected Error or Warning
                  Status: None
             Assigned to: None
         Originator Name: 
        Originator Email: 
             Open/Closed: Open
                 Release: 9.0.90
         Discussion Lock: Any
        Operating System: Any
           Fixed Release: None
         Planned Release: None


    _______________________________________________________

Follow-up Comments:


-------------------------------------------------------
Date: Sat 02 Mar 2024 11:15:29 PM EST By: Nicholas Jankowski <nrjank>
see https://octave.discourse.group/t/mad-in-9-0-91/5340
and https://octave.discourse.group/t/pkg-test-package-fails/5160

pkg test statistics is failing with v1.6.4 and octave 9.0.91.  some of the
tests appear to be because certain function that are only supposed to be
shadowed for octave <v9 are being tested by octave 9.0.91.  it _seems_ that
the mad.m tests from the package version are being tested by mad.m from octave
core.  this appears most obvious with the input validation checks, where 'pkg
test statistics' is throwing a fail because the error checks are failing with
different error messages.


***** error <Invalid call> mad (1, 2, 3, 4)
!!!!! error failed.
Expected <Invalid call>, but got <mad: function called with too many inputs>

***** error <mad: X must be numeric.> mad ("text")
!!!!! error failed.
Expected <mad: X must be numeric.>, but got <mad: X must be a numeric vector
or matrix>

***** error <mad: X must be numeric.> mad ({2 3 4})
!!!!! error failed.
Expected <mad: X must be numeric.>, but got <mad: X must be a numeric vector
or matrix>

***** error <mad: FLAG must be either 0 or 1.> mad (1, "all", 3)
!!!!! error failed.
Expected <mad: FLAG must be either 0 or 1.>, but got <mad: OPT must be 0 or
1>

***** error <mad: FLAG must be either 0 or 1.> mad (1, "b")
!!!!! error failed.
Expected <mad: FLAG must be either 0 or 1.>, but got <mad: OPT must be 0 or
1>


The first case occurs because octave started automating the 'too many inputs'
failures, but the shadowed version kept this test for older versions of
octave.

running pkg test statistics returns:


..\mingw64\share\octave\packages\statistics-1.6.4\shadow9\mad.m  pass   74/82
                                                                   FAIL    8


while loading the package shows that the file isn't being shadowed, and it
runs a test against the core function:


>> pkg load statistics
'mad' is a function from the file
C:\devtools\Octave\octave-stable\mingw64\share\octave\9.0.91\m\statistics\mad.m
>> test mad
PASSES 61 out of 61 tests


is this intended behavior? is it occurring because of a pkg issue or is there
an issue with how the shadowing rules are defined in the package (it's set for
9.1 and 9.0.91 confuses it? not sure where to check for that.)










    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?65406>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/