Kraken updates

[EricTRocks]

This PR ports changed done locally in Kraken to pyflowgraph to extend the functionality and to allow Kraken to make pyflowgraph a sub-module.

[EricTRocks]

@willjp @drogenlied Could you have a look to see if there is anything that stands out as incompatible with PySide2 / Python3

[drogenlied]

I tested pyqt4 and pyqt5 on Python3 and there are definitely some issues:

• ./setup.py develop does not lead to a working dev installation, all the python files except for setup.* need to be in a folder named pyflowgraph
• There is a python2/3 compatibility issue issue here, check the iteritems entry on http://python-future.org/compatible_idioms.html or line 295 in graph_view.py for solutions:

  Traceback (most recent call last):
  File "/home/jan/build/pyflowgraph_eric/pyflowgraph/graph_view.py", line 461, in mouseMoveEvent
  for name, node in self.__nodes.iteritems():
  AttributeError: 'dict' object has no attribute 'iteritems'

• There is an undefined variable here, no clue what is going on:

  Traceback (most recent call last):
  File "/home/jan/build/pyflowgraph_eric/pyflowgraph/graph_view.py", line 491, in mouseMoveEvent
  delta = event.pos() - self._lastMousePos
  AttributeError: 'GraphView' object has no attribute '_lastMousePos'

• The structure below has several problems, with pyqt5, QtCore.Qt.MouseButton.MiddleButton and QtCore.Qt.MouseButton.RightButton do not exist and the is comparison has caused problems for me previously, so i would recommend comparing with ==:

  if event.button() is QtCore.Qt.LeftButton:
  ...
  elif event.button() is QtCore.Qt.MouseButton.MiddleButton:
  ...
  elif event.button() is QtCore.Qt.MouseButton.RightButton:
  ...

• Both tests have a segmentation fault after closing the window.
• Moving nodes with connections leaves artifacts from the connection line, this can probably be fixed by requesting a redraw of the union of the old and new bounding rectangles of the connections.

[EricTRocks]

@drogenlied Thanks for testing. The problem with keeping the nested pyflowgraph directory is that it can no longer be easily included as a submodule in other git projects where it's cloned location in that master project is within a path that is automatically added to PYTHONPATH. I tested normal installation with pip and it worked. What are you referring to when you are saying a dev installation?

I'll take a look at the rest of the issues and fix them up.

[EricTRocks]

@drogenlied I did a setup.py develop on my end and it created an egg-link file and I could then import pyflowgraph without issue. What is not working for you? regarding thing?

[drogenlied]

@EricTRocks, when I use setup.py develop, it creates the link just fine but I can't import pyflowgraph. This happens on both Python 2.7.13 and 3.5.3. However, I can do import graph_view anywhere, which then fails because it cannot import Node from the relative path .node.

[EricTRocks]

@drogenlied OK thanks for that information. I've got Python 2.7.9 installed I believe. Will upgrade and see if I get the same issues.

[EricTRocks]

@drogenlied I actually did get the error with the develop mode install. Just needed to restart the command prompt to see it. I've fixed all the issues you found previously I believe with the recent commits aside from the connection line artifacts.

I've logged another issue for the artifacts. https://github.com/EricTRocks/pyflowgraph/issues/9

Let me know how the recent changes work for you.

[EricTRocks]

@drogenlied The artifacts for connections is fixed (#9) with the latest commit. This was due to a change that was made where in graph_view.py it was calling node.moveBy() instead of node.translate(). The latter is a custom method which calls prepareConnectionGeometryChange() that is used to alert the system that the connections need to be redrawn.

[drogenlied]

Things look ok now, except for the Python3/pyqt4 segfault on exit, which may not even be reproducible with slightly different versions of Qt or other operating systems.

[EricTRocks]

I'll open a ticket for the segfault. In the mean time I'm merging this in.