michaelhush/M-LOOP

Updated for the git tag update

setup.py test was failing on windows machines because if
__name__==__main__ was not included in the setup.py folder. This has
been fixed. Also updated setup.py to explicitly include all package with
version requirements to help get it working on Travis CI.

Added install for dependency issues

Added configuration file

Update learners file. Missed in last update.

Queue listener and logger have been removed as they were causing issues
on Windows and python 2. Logging in a process is now direct to the cmd
line. There are no duplicate loggers anymore.

Documentation has been updated to reflect the support of python 2 and
3.

Also deleted some code which treated the interface as a multiprocessing
class.

Updated the package so it now works on python 2 and 3. All use of super
had to be updated and compatibility headers were added to all module

The logger was causing problems with multiprocessing on windows. Windows
spawns a new python thread then pickles the process object. A logger can
not be pickled. The solution was to delete the logger at the end of init
then recreate at the start of run.

Removed step from install instructions on installing scikit-learn as
the source is now already included.

the required code from scikit-learn is now included in mloop. When
scikit-learn releases the latest dev version, this code will be removed
and mloop will require scikit-learn as a dependency. But this is a
better solution than requiring installation of scikit-learn from source
in the short term.

Interface is no longer run as a seperate process but just a seperate
thread. This is to improve compatibility with matplotlib on MacOS and
Windows.

Hard coded some versions

Completed documentation.

Fixed a few bugs with file writing formats with ‘txt’ in particular

Attempting to get docs working on read the docs

Fixed version tag

Added another example for the tutoiral, and made a corresponding test.

Fixed bug with reading matlab files for plots.

Changed some keywords to do with file_types and files.

Moved version to 2.0.1

Complete rewriting of M-LOOP to make it more extensible and improve
performance.

M-LOOP is now includes threading and multiprocessing to remove any
unnecessary delays between experiments.

M-LOOP is installable with setuptool.

M-LOOP now had documentation generated with sphinx.