dump/restore methods exposing writeGraphModel/readGraphModel

[prusse-martin]

No description provided.

[nicoddemus]

Lint has failed:

qmxgraph/api.py:374:80: E501 line too long (83 > 79 characters)

Could you please fix that and amend?

[coveralls]

Coverage decreased (-0.3%) to 88.667% when pulling 5f54f4b on fb-ASIM-198-project-file-io into 3f9265a on master.

[nicoddemus]

Waiting for @fogo to give his blessing. 👍

[prusse-martin]

Did the lint fix.
The coverage got down since the 2 functions added are being tested only using selenium to check the actual javascript implementation. Do you think it is worth to have a test covering the wrapper functions in api.py?

[nicoddemus]

Let's see what @fogo has to say about it.

[fogo]

The new feature seems fine. It seems this is the expected way to store/restore a mxGraph. Just take on future changes to make sure they are always compatible with this (i.e. they are visually identical after restored).

About the coverage on API, this is already a problem actually. If you take a deeper look, you'll notice that there're other methods uncovered. I guess it is time to decide to either cover them (more work, safer) all or maybe just don't cover it at all (risky, API isn't always straightforward between Python/Qt/JS).

[prusse-martin]

Sorry for delaying the comment.
@nicoddemus think it is better to have coverage on this and I kind agree with him.
He suggested at least have tests mocking the actual javascript functions so just "API" is tested (the actual javascript is already tested).

I can add that later. If you think mocking is good enough I plan to use testdouble.js or Sinon.JS.

[fogo]

Sorry for the late answer, but mocking feels like a good start. I look forward to the future PR ;)