[WIP] Steppath and Line2D

[Tillsten]

Ok, this is a first try to refactor Lines2D for steps. Submitted as a PR to run the tests. See #1709.

[Tillsten]

So tests are still passing, any other concerns with this approach?

[Tillsten]

Ok, because i not sure if it is worth to continue this PR i am closing this.
I am still thinking that refactoring out the path calculation is the right think to do, but i am not so sure about the API change. Maybe just adding a get_drawn_path() is a better way.

[mdboom]

@Tillsten: Other than the lack of response (for which I apologise -- the PR queue can be a real firehose sometimes), what's your reason for closing this? It seems reasonable...

[Tillsten]

@mdboom Mostly because i not sure if this is the right way to do it: the idea of the old design seems to be that the difference between draw_styles is only relevant for drawing.

The biggest practical concerns i have, is that i am not sure if line.set_data still works as as expected. The fact the test-suit works seems to imply it does, but i am still unwary. The return value of get_data now returns also something different (probably more corrected now), which is the aim of the PR.

[tacaswell]

I would put a vote in for not having the function change the data so that set_[x,y]data/get_[x,y]data round trip works. Aside from my sense of symmetry, changing the data would mess up animation code where you append new data onto existing lines (by using get_[x,y]data -> append -> set_[x,y]data).

[Tillsten]

@tacaswell But the problem still remains, that at the moment you don't have any method to get the drawn data. Also the step-vertices are recalculated very draw.

[tacaswell]

@Tillsten What do you mean by drawn data?

Is the calculation too slow to run every draw?

[Tillsten]

@tacaswell The problem that there is no way of getting the "real" vertices from a path which use steps. This problematic if you want to make a filled histogram. Speedwise: I not sure, i don't use mpl for a lot of interactive work because it is quite slow, so i didn't benchmark it. I think the difference will be only visible for a lot of step-plots.

[tacaswell]

@Tillsten If you assume a bottom level, can you put in an axhline and then use fill_between?

[thriveth]

@tacaswell No, see the link in the OP here: #1709 (it is old but still applies).