Matlab strings--potential fix for scipy read error

[dcwhite]

@jessdtate can you verify this fixes the scipy error referenced in #1692? Also check round-trip save/load for basically every other matlab file type and program that reads/writes such types...

[jessdtate]

There seems to be a problem reading integer matrices into scirun now. Before, they could be read from matlab. Here is the error from a saved matrix (same from Python and matlab):

Message: Import failed.
ERROR: /Users/jess/software/SCIRun/src/Modules/DataIO/GenericReader.h(190): Throw in function virtual void SCIRun::Modules::DataIO::GenericReader >, SCIRun::Modules::MatrixPortTag>::execute() [HType = boost::shared_ptr >, PortTag = SCIRun::Modules::MatrixPortTag] Dynamic exception type: boost::exception_detail::clone_impl std::exception::what: Import failed. [SCIRun::Core::tag_error_message*] = Import failed.

Matlab_save_int_matrix.mat.zip
python_save_int_matrix.mat.zip

[jessdtate]

There is a bit of inconsistency with ints in matrices, which may be the source of the issue. it seems the modules that work (like getfielddata) just convert it to a double. This doesn't seem to be happening correctly in read field.

[dcwhite]

OK I'll take a look

[jessdtate]

a slight correction, ints from matlab didn't work either, but the error was just: import failed.

[jessdtate]

I can get it to work with any fields. I still get the error: TypeError: Expecting miINT8 as data type

[dcwhite]

OK I'll close this and come up with a different fix.

[jessdtate]

Maybe we should merge this into master? It works with matrix files, so it's half fix at least. I've been including it in local builds for a while to deal with matrix files.

[dcwhite]

I can see if I can merge it in with a feature flag--enabled at run-time, or a setting which you can save as turned on.