bugGNU Octave - Bugs: bug #53636, failing test under Windows:...

 
 

bug #53636: failing test under Windows: bsxfun.cc-tst

Submitter:  Hartmut <hardy>
Submitted:  Fri 13 Apr 2018 11:53:27 AM UTC
   
 
Category:  Test Suite Severity:  3 - Normal
Priority:  5 - Normal Item Group:  Regression
Status:  Duplicate Assigned to:  None
Originator Name:  Open/Closed:  * Closed
Release:  * 4.3.90 Operating System:  * Microsoft Windows
Fixed Release:  None Planned Release:  None
* Mandatory Fields

Add a New Comment Rich Markup
   

Jump to the original submission

Sat 14 Apr 2018 12:17:57 PM UTC, comment #12: 

I pushed the temporary fix here http://hg.savannah.gnu.org/hgweb/octave/rev/9b93b85459da.  Marking this bug as a duplicate and closing report.

Rik <rik5>
Group administrator
Sat 14 Apr 2018 11:31:06 AM UTC, comment #11: 

I have tested on a Win7 machine (as I have currently no access to the Win10 machine from comment #0):

  • running the loop code from comment #7 on the original bsxfun.cc-tst (included 4.3.90) does also throw an error once in a while.
  • running the modified bsxfun.cc-tst file from file #43920 works totally fine. (loop result=200)


So I would say that this modified test file is appropriate for a temporary workaround as proposed in comment #5).

Hartmut <hardy>
Fri 13 Apr 2018 10:01:36 PM UTC, comment #10: 

@Dmitri: Was your testing on Windows or a Linux machine?

Rik <rik5>
Group administrator
Fri 13 Apr 2018 09:51:05 PM UTC, comment #9: 

Works for me too.

Dmitri.

Dmitri A. Sergatskov <dasergatskov>
Fri 13 Apr 2018 08:52:55 PM UTC, comment #8: 

I got 200 (on Debian).

Would be good to get a verification from Hartmut or from someone else on Windows with 4.3.90.

Mike Miller <mtmiller>
Group Member
Fri 13 Apr 2018 08:00:42 PM UTC, comment #7: 

I have made a preliminary changeset to fix the rand and randn random number generator to a known state before the BIST tests are done.  However,m s I never see a probleo I need someone else to help do the testing.

I am uploading a file bsxfun.cc-tst.  To test, download this file and place it in the current directory where you start Octave (so it is picked before the default version).  Then run


bm = zeros (200,1);
for i = 1:200
  bm(i) = test ("bsxfun.cc-tst");
endfor
sum (bm)


If it passes (after about 15 minutes) then we are good.  Otherwise, adjust the random number generator state to a different value (currently 42) and re-run.




(file #43920)

Rik <rik5>
Group administrator
Fri 13 Apr 2018 07:28:12 PM UTC, comment #6: 

Good plan.

Mike Miller <mtmiller>
Group Member
Fri 13 Apr 2018 05:00:28 PM UTC, comment #5: 

I propose that we

1) Fix the random seed ahead of these tests to a known good value (temporary workaround)
2) Close this bug report as a duplicate of bug #51779
3) Take up the questions in bug #51779 on the development branch

Thoughts?

Rik <rik5>
Group administrator
Fri 13 Apr 2018 04:10:23 PM UTC, comment #4: 

This is similar and may be the same as bug #51779.

That bug fails for me (and on buildbots) occasionally because bsxfun with @pow is doing a different operation between float operands than the test was written for.

This intersects with the larger overall bug that Matlab does mixed-type operations in double and casts the result, while Octave does mixed-type operations in single.

Mike Miller <mtmiller>
Group Member
Fri 13 Apr 2018 04:08:20 PM UTC, comment #3: 

I got 1 failure out of 100.

I am pretty sure it is the same as #51779

Dmitri.
--

Dmitri A. Sergatskov <dasergatskov>
Fri 13 Apr 2018 03:29:49 PM UTC, comment #2: 

I tried running the test 100 times and it works for me


for i = 1:100
  bm(i) = test ('bsxfun.cc-tst');
endfor
sum (bm)



Rik <rik5>
Group administrator
Fri 13 Apr 2018 03:07:11 PM UTC, comment #1: 

The test file is mentioned in fntests.log


>>>>> processing C:\Octave\OCTAVE~1.90\share\octave\4.3.90\etc\tests\libinterp\corefcn\bsxfun.cc-tst


Copy that file to your current working directory so that it is on the search path for Octave.  Then do


test bsxfun.cc-tst


Also, is this related to bug #51779?

Rik <rik5>
Group administrator
Fri 13 Apr 2018 11:53:27 AM UTC, original submission:  

I installed the official 4.3.90 w64 installer under Win 10.

One (and only one) test is failing: bsxfun.cc-tst . This test was not yet failing with Octave 4.2.2 on the same PC.

This is what I found in fntest.log :


>>>>> processing C:\Octave\OCTAVE~1.90\share\octave\4.3.90\etc\tests\libinterp\corefcn\bsxfun.cc-tst
***** test
 funs = {@plus, @minus, @times, @rdivide, @ldivide, @power, @max, @min, ...
         @rem, @mod, @atan2, @hypot, @eq, @ne, @lt, @le, @gt, @ge, ...
         @and, @or, @xor };

 float_types = {@single, @double};
 int_types = {@int8, @int16, @int32, @int64, ...
              @uint8, @uint16, @uint32, @uint64};

 x = rand (3) * 10-5;
 y = rand (3,1) * 10-5;

 for i=1:length (funs)
   for j = 1:length (float_types)
     for k = 1:length (int_types)

       fun = funs{i};
       f_type = float_types{j};
       i_type = int_types{k};

         assert (bsxfun (fun, f_type (x), i_type (y)), ...
                 fun (f_type(x), i_type (y)));
         assert (bsxfun (fun, f_type (y), i_type (x)), ...
                 fun (f_type(y), i_type (x)));

         assert (bsxfun (fun, i_type (x), i_type (y)), ...
                 fun (i_type (x), i_type (y)));
         assert (bsxfun (fun, i_type (y), i_type (x)), ...
                 fun (i_type (y), i_type (x)));

         assert (bsxfun (fun, f_type (x), f_type (y)), ...
                 fun (f_type (x), f_type (y)));
         assert (bsxfun (fun, f_type(y), f_type(x)), ...
                 fun (f_type (y), f_type (x)));
     endfor
   endfor
 endfor
!!!!! test failed
ASSERT errors for:  assert (bsxfun (fun, f_type (x), i_type (y)),fun (f_type (x), i_type (y)))

  Location  |  Observed  |  Expected  |  Reason
   (2,2)       5442282      5442283      Abs err 1 exceeds tol 0 by 1
shared variables   scalar structure containing the fields:

    a =

    ans(:,:,1) =

       1.78672
       0.27992
      -0.61915

    ans(:,:,2) =

      -0.25891
       1.30055
      -0.49092

    ans(:,:,3) =

      -0.098843
      -0.482358
      -0.724215

    b =

    ans(:,:,1,1) =

       0.052885   0.068687   0.665253

    ans(:,:,2,1) =

       1.74760  -0.87201  -0.85146

    ans(:,:,3,1) =

      -0.701998   0.198724   0.093790

    ans(:,:,1,2) =

       1.88442  -1.05123  -0.43862

    ans(:,:,2,2) =

       0.33925  -0.66327   1.39562

    ans(:,:,3,2) =

      -0.31050  -0.51481   1.57162

    ans(:,:,1,3) =

      -0.106352   0.929991  -0.081016

    ans(:,:,2,3) =

      -1.40578  -1.72386   0.12906

    ans(:,:,3,3) =

       0.33246  -0.13230   1.02211

    aa =

    ans(:,:,1,1) =

       1.78672   1.78672   1.78672
       0.27992   0.27992   0.27992
      -0.61915  -0.61915  -0.61915

    ans(:,:,2,1) =

      -0.25891  -0.25891  -0.25891
       1.30055   1.30055   1.30055
      -0.49092  -0.49092  -0.49092

    ans(:,:,3,1) =

      -0.098843  -0.098843  -0.098843
      -0.482358  -0.482358  -0.482358
      -0.724215  -0.724215  -0.724215

    ans(:,:,1,2) =

       1.78672   1.78672   1.78672
       0.27992   0.27992   0.27992
      -0.61915  -0.61915  -0.61915

    ans(:,:,2,2) =

      -0.25891  -0.25891  -0.25891
       1.30055   1.30055   1.30055
      -0.49092  -0.49092  -0.49092

    ans(:,:,3,2) =

      -0.098843  -0.098843  -0.098843
      -0.482358  -0.482358  -0.482358
      -0.724215  -0.724215  -0.724215

    ans(:,:,1,3) =

       1.78672   1.78672   1.78672
       0.27992   0.27992   0.27992
      -0.61915  -0.61915  -0.61915

    ans(:,:,2,3) =

      -0.25891  -0.25891  -0.25891
       1.30055   1.30055   1.30055
      -0.49092  -0.49092  -0.49092

    ans(:,:,3,3) =

      -0.098843  -0.098843  -0.098843
      -0.482358  -0.482358  -0.482358
      -0.724215  -0.724215  -0.724215

    bb =

    ans(:,:,1,1) =

       0.052885   0.068687   0.665253
       0.052885   0.068687   0.665253
       0.052885   0.068687   0.665253

    ans(:,:,2,1) =

       1.74760  -0.87201  -0.85146
       1.74760  -0.87201  -0.85146
       1.74760  -0.87201  -0.85146

    ans(:,:,3,1) =

      -0.701998   0.198724   0.093790
      -0.701998   0.198724   0.093790
      -0.701998   0.198724   0.093790

    ans(:,:,1,2) =

       1.88442  -1.05123  -0.43862
       1.88442  -1.05123  -0.43862
       1.88442  -1.05123  -0.43862

    ans(:,:,2,2) =

       0.33925  -0.66327   1.39562
       0.33925  -0.66327   1.39562
       0.33925  -0.66327   1.39562

    ans(:,:,3,2) =

      -0.31050  -0.51481   1.57162
      -0.31050  -0.51481   1.57162
      -0.31050  -0.51481   1.57162

    ans(:,:,1,3) =

      -0.106352   0.929991  -0.081016
      -0.106352   0.929991  -0.081016
      -0.106352   0.929991  -0.081016

    ans(:,:,2,3) =

      -1.40578  -1.72386   0.12906
      -1.40578  -1.72386   0.12906
      -1.40578  -1.72386   0.12906

    ans(:,:,3,3) =

       0.33246  -0.13230   1.02211
       0.33246  -0.13230   1.02211
       0.33246  -0.13230   1.02211


How do I repeat a cc-tst test? Just "test bsxfun.cc-tst" did not work.

Hartmut <hardy>

 

(Note: upload size limit is set to 16384 kB, after insertion of the required escape characters.)

Attach Files:
   
   
Comment:
   

Attached Files
file #43920:  bsxfun.cc-tst added by rik5 (6KiB - application/octet-stream)

 

Digest:
   bug dependencies.

Items that depend on this one: None found

 

Carbon-Copy List
  • -email is unavailable- added by mtmiller (Posted a comment)
  • -email is unavailable- added by dasergatskov (Posted a comment)
  • -email is unavailable- added by rik5 (Posted a comment)
  • -email is unavailable- added by hardy (Submitted the item)
  •  

    There are 0 votes so far. Votes easily highlight which items people would like to see resolved in priority, independently of the priority of the item set by tracker managers.

    Only group members can vote.

     

    Follow 5 latest changes.

    Date Changed by Updated Field Previous Value => Replaced by
    2018-04-14 rik5 StatusNeed Info Duplicate
        Open/ClosedOpen Closed
    2018-04-13 rik5 Attached File- Added bsxfun.cc-tst, #43920
    2018-04-13 mtmiller Dependencies- Depends on bugs #51779
    2018-04-13 rik5 StatusNone Need Info

    Back to the top

    Powered by Savane 3.13-4448.
    Corresponding source code