Multifigure toolbar

[fariza]

Getting rid of the canvas property inside FigureManager and replacing it with a figure property with setter and getter
Also, allows to instantiate FigureManager without any figure.
In window added a method to replace_element this allows to reuse the same FigureManager object with different figures.

[fariza]

Why this PR is showing me 38 commits? when I only made one?

[OceanWolf]

Hmm, not sure when you pulled from my branch, perhaps I had done a merge/rebase, in which case you need to update your local copy of my branch and then do a rebase.

[fariza]

But, in general what do you think of this idea?

[OceanWolf]

Can you clear up your PR first (rebase) so that I can take a good look.

[fariza]

There is nothing to rebase, your PR (From your branch to MPL master) hasn't changed

[fariza]

I rebased again from the your PR on MPL.
Still I see too many commits here, I don't get it.

[OceanWolf]

@fariza that looks slightly better, it removes some of my commits from your PR, but still very odd.

You have commits from master since my last rebase (no idea how they got in there), and I miss commits prior to my last rebase (no idea why I don't have them).

[fariza]

If you check fariza@07d30a6 you'll see the main differences.

I will make a clean PR if you agree with the way this one is going. In any case I'll change a couple of minor things. But before putting more time into it, I would like to have your opinion

[OceanWolf]

I just had a brainwave as to what could have caused this... did you branch this from mine, or did you branch this say from master and then later rebased it from my branch?

[fariza]

I branched this from the PR on MPL,

[OceanWolf]

Hmm, that might explain it... when I made PRs against MEP22 I branched from your repo rather than the PR... I will look up how to do it.

[OceanWolf]

Ahh yes, it comes back to me, I did it in the reverse order (cloning from you and then later from matplotlib), but basically you have to add a new remote to your local repo.

From the command line I get:

$ git remote -v
master https://github.com/matplotlib/matplotlib.git (fetch)
master https://github.com/matplotlib/matplotlib.git (push)
origin https://github.com/OceanWolf/matplotlib.git (fetch)
origin https://github.com/OceanWolf/matplotlib.git (push)
upstream https://github.com/fariza/matplotlib.git (fetch)
upstream https://github.com/fariza/matplotlib.git (push)

If I had done it in the order you should do, then you would have upstream pointing to matplotlib (instead of mine pointing to yours), and then you can name the remote pointing to mine as oceanwolf, that way you sync your backend-refactor branch against my fork, rather than the PR on mpl. Make sense?

[OceanWolf]

Ofc when you push, you push to origin and then make a PR on github, but for fetching it should come from a remote pointing to my fork.