Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Hard transition to Python3 as a squashed merge (#231)
* futurize stage1, tests pass * Catching two exceptions * futurize stage2, tests fail * Took out python2 wrappers, excessive casts, and fixed small things along the way * Some old indentation errors * Adapted to API change in configparser * Fixed programmatic imports * Removed explicit checks for iteritems() * Adapted to api change in socket * Removed all remaining python2 wrappers * avoiding configparser DuplicationErrors more generally * pickle doesn't do local classes like cPickle did * Missed a few old_div references * Found some remaining subscripting of iterators * Fixed two PtyScan classes - you have to be careful with /= on arrays in python 3 * Fixed the live plotter * warn() method deprecated in logging * imp module is deprecated * Enough with the 'Failed to register' warnings * Removed pickling unsupported data types from h5io * Update travis.yaml, full_dependencies and core dependencies for python 3 ` * fixes saving of np.recarrays by casting to fixed length string before saving. * Reinstates pickling of unknown types. * Update doc conf to python3 * Refactor of script2rst.py gets rid of 1800 or so errors. * Corrects for new lines in the middle of a sentence in .help. * The .scan field doesn't exist anymore * converts ptypy.inspect to python3 * Move __idiv__ --> __truediv__ * Needs an escape character * extra line here * removes invalid doc strings. The class doc string already contains the necessary parameters anyway * removes invalid, duplicated docstring. * fixes test that fails because I changed to __truediv__. It would have been good if this had failed on implementation rather than checking it was there. * backslashes aren't needed here * Turns autoplot on for this. * add path to the data * The cxro lookup was a victim of the python3 convertor. Works now. * This isn't an integer in p3. It needs to be * Defaults tree has moved * Need to check high and low limits properly for iterables too * Needs to be u.Param * Allows backwards compatibility. We can now h5read files that were saved in the python2.7 version of ptypy * removed record array cause it was not used as such. Now it's a simple structured array where unicode is replaced with byte strings. h5py can not handle numpy U dtypes in arrays * Bumped to version 0.4.0 in preparation for merge * Updated release notes * bringing the doc up to right version automatically
- Loading branch information