test_bfopen fails with

[sbesson]

While trying to create a minimal example file to reproduce https://forum.image.sc/t/bio-formats-octave-error-on-bfsave/27504, I realized that the current Travis build is failing - see https://travis-ci.org/ome/bio-formats-octave-docker/builds/545619855.

Populating metadata
Reading series #1
    .error: octave_base_value::reshape (): wrong type argument 'octave_java'
error: called from
    bfGetPlane at line 99 column 7
    bfopen at line 150 column 13
    test_bfopen at line 6 column 6
    travis_test at line 9 column 1

There are two issues associated with this error:

• I would have expected the docker run command to fail with a non-zero status code on a failing script rather than a false negative
• octave_base_value::reshape (): wrong type argument 'octave_java' is thrown when calling bfGetPlane for all recent builds using GNU Octave 4.2 / Bio-Formats 6. The only related report I could find was https://savannah.gnu.org/bugs/?44881. @carandraug: do you have any immediate thoughts on this before we start investigating.

[carandraug]

I would have expected the docker run command to fail with a non-zero status code on a failing script rather than a false negative

I've checked on Octave 4.0 and Octave 6.0 and in both cases returns a non zero return code

$ touch "test&sizeT=3.fake"
$ octave travis_test.m "test&sizeT=3.fake"
[...]
s = 1
v = 6.2.0
FakeReader initializing test&sizeT=3.fake
Reading series #1
    .error: octave_base_value::reshape (): wrong type argument 'octave_java'
error: called from
    bfGetPlane at line 99 column 7
    bfopen at line 150 column 13
    test_bfopen at line 6 column 6
    travis_test at line 9 column 1
$ echo $?
1

octave_base_value::reshape (): wrong type argument 'octave_java' is thrown when calling bfGetPlane for all recent builds using GNU Octave 4.2 / Bio-Formats 6. The only related report I could find was https://savannah.gnu.org/bugs/?44881.

That issue is unrelated (the issue is about making reshape also work in java arrays and that is something that doesn't even work in Matlab). This error seems to be related to auto boxing/unboxing rules. In Matlab, it looks like I in:

I = javaMethod('makeDataArray2D', 'loci.common.DataTools', plane, ...
    bpp, fp, little, ip.Results.height);

returns a Matlab array of class int8 but in Octave this remains an octave_java object of class byte[][] which is why the switch statement fails. I believe that in Octave, arrays with more than one dimension remain a java object in case they are non-rectangular. This probably needs to be changed in Octave for sake of Matlab compatibility but I don't think there's anyone working in Octave at the moment that quite understands the different conversion rules that would be Matlab compatible.

[sbesson]

@carandraug: thanks for the quick reponse. The base image used for testing Octave is Ubuntu 18.04 which ships GNU Octave 4.2.2

octave@89da10fab560:~$ touch "test&sizeT=3.fake" 
octave@89da10fab560:~$ octave -version
octave: X11 DISPLAY environment variable not set
octave: disabling GUI features
GNU Octave, version 4.2.2
Copyright (C) 2018 John W. Eaton and others.
This is free software; see the source code for copying conditions.
There is ABSOLUTELY NO WARRANTY; not even for MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE.

Octave was configured for "x86_64-pc-linux-gnu".

Additional information about Octave is available at http://www.octave.org.

Please contribute if you find this software useful.
For more information, visit http://www.octave.org/get-involved.html

Read http://www.octave.org/bugs.html to learn how to submit bug reports.
octave@89da10fab560:~$ octave travis_test.m "test&sizeT=3.fake" && echo passed
octave: X11 DISPLAY environment variable not set
octave: disabling GUI features
s = 1
v = 6.1.0
FakeReader initializing test&sizeT=3.fake
Reading series #1
    .error: octave_base_value::reshape (): wrong type argument 'octave_java'
error: called from
    bfGetPlane at line 99 column 7
    bfopen at line 150 column 13
    test_bfopen at line 6 column 6
    travis_test at line 9 column 1
passed
octave@89da10fab560:~$ echo $?
0
octave@89da10fab560:~$ exit

Switching to Ubuntu 16.04 and GNU Octave 4.0.0, I can confirm the status code is returned as non-zero as expected

octave@50c2a3c49946:~$ touch "test&sizeT=3.fake" 
octave@50c2a3c49946:~$ octave --version
octave: X11 DISPLAY environment variable not set
octave: disabling GUI features
GNU Octave, version 4.0.0
Copyright (C) 2015 John W. Eaton and others.
This is free software; see the source code for copying conditions.
There is ABSOLUTELY NO WARRANTY; not even for MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE.

Octave was configured for "x86_64-pc-linux-gnu".

Additional information about Octave is available at http://www.octave.org.

Please contribute if you find this software useful.
For more information, visit http://www.octave.org/get-involved.html

Read http://www.octave.org/bugs.html to learn how to submit bug reports.
octave@50c2a3c49946:~$ octave travis_test.m "test&sizeT=3.fake"
octave: X11 DISPLAY environment variable not set
octave: disabling GUI features
s =  1
v = 6.1.0
FakeReader initializing test&sizeT=3.fake
Reading series #1
    .error: octave_base_value::reshape (): wrong type argument 'octave_java'
error: called from
    bfGetPlane at line 99 column 7
    bfopen at line 150 column 13
    test_bfopen at line 6 column 6
    travis_test.m at line 9 column 1
octave@50c2a3c49946:~$ echo $?
1

This seems related to http://octave.1599824.n4.nabble.com/Octave-s-exit-status-td4680966.html. It looks like the regression was fixed in 6.0 but did not make it into a patch release of Octave 4.2.x.

Thanks also for pointing at the autoboxing behavior. While bisecting changes introduced, I think this is effectively a regression introduced as part of ome/bioformats#3301.

[jburel]

The same problem occurs when testing on docker image Ubuntu 22.04 and octave 6.4.0

[carandraug]

I've looked at this and it works fine in bioformats 5.9.2 and started to fail in bioformats 6.0.0. @sbesson suggested and I confirm that the issue is a regression introduced in ome/bioformats#3301

It boils down to the description I had above:

This error seems to be related to auto boxing/unboxing rules. In Matlab, it looks like I in:

I = javaMethod('makeDataArray2D', 'loci.common.DataTools', plane, ...
    bpp, fp, little, ip.Results.height);

returns a Matlab array of class int8 but in Octave this remains an octave_java object of class byte[][] which is why the switch statement fails. I believe that in Octave, arrays with more than one dimension remain a java object in case they are non-rectangular. This probably needs to be changed in Octave for sake of Matlab compatibility but I don't think there's anyone working in Octave at the moment that quite understands the different conversion rules that would be Matlab compatible.

I will prepare a pull request.

[carandraug]

I will prepare a pull request.

Done in ome/bioformats#4130