Test suite broken

[jaeandersson]

The test suite got (more) broken in one of the more recent commits.

[jaeandersson]

What is causing this is that there is no good way of disabling all the tests that are known not to work. At least in my understanding. For all practical purposes, the test suite has always been broken for MATLAB. Sometimes changes to the module will cause some tests to fail more violently than before (e.g. crashing MATLAB), but that should not be a reason to hold back a commit.

[jaeandersson]

starting a git bisect

[jaeandersson]

The violently failing example appesrs to be dynamic_cast

[jaeandersson]

jaeandersson@jaeandersson-VirtualBox:~/dev/swig$ git bisect log git bisect start

bad: [98ed1086ac8d292a741e92fb4820efa3fc0d5991] Merge branch 'dcb210-matlab' into matlab

git bisect bad 98ed1086ac8d292a741e92fb4820efa3fc0d5991

good: [8e13c91b0523a27dda9750406f672f7fe615d19a] Revert "[MATLAB] issue #47: Fixes 'clear all' bug"

git bisect good 8e13c91b0523a27dda9750406f672f7fe615d19a

good: [8f4101ff1f64d1cb397416ee4680a7fa9ab8b1a2] [MATLAB] issue #44: Enforced modulenameMATLAB_wrap naming convension for compiled mex function

git bisect good 8f4101ff1f64d1cb397416ee4680a7fa9ab8b1a2

bad: [b2aee0e8297ded4fa27e95dc48c2698ac8148ebe] [MATLAB] issue #44: Removed unused variables

git bisect bad b2aee0e8297ded4fa27e95dc48c2698ac8148ebe

bad: [f1a015516197cdc8c2e2cea484ee226ac526e919] [MATLAB] issue #44: Removed setup script functionality Usually not needed and when needed, any script can be called from the init function

git bisect bad f1a015516197cdc8c2e2cea484ee226ac526e919

bad: [90e792fd10a010f73bd44daa07d4fdbd0b4b54ec] [MATLAB] issue #44: Refactored initialization of modules

git bisect bad 90e792fd10a010f73bd44daa07d4fdbd0b4b54ec

good: [72c5ccb20c30e049fa4426566211500534eeca5c] [MATLAB] issue #44: Added option for specifying mex-function infix

git bisect good 72c5ccb20c30e049fa4426566211500534eeca5c

first bad commit: [90e792fd10a010f73bd44daa07d4fdbd0b4b54ec] [MATLAB] issue #44: Refactored initialization of modules

[KrisThielemans]

This is outdated, certainly after #96. I suggest to close this