added fill_percent to params for insert, swap and remove

[Kully]

This PR accomplishes:

• added fill_percent to params for insert, swap and remove
• fix HTML preview to reflect fill_percent
• update changelog
• remove #TODOs from dashboard_objs.py

re: #711

[Kully]

@cldougl ready for a 👀

[cldougl]

hey @Kully this looks good, I'm trying out some examples now! Are these changes intended to be backwards compatible?

[cldougl]

I would say this is more of added functionality than a bug fix (do you agree?) so I think it should be version 2.5.0 based on: https://semver.org/

[Kully]

I agree they are not bug fixes. The question I have now about my message below is if we consider these points as backwards-compatible. Functionally the API is the same and produces dashboards that look the same. I just had to modify a test or two.

[cldougl]

Is there no way to add this functionality as an option rather than the default so that it remains backwards compatible?

[Kully]

I don't think it's worth it. There was a technical issue with, before, using pixels as the metric for slicing up the boxes (since it was always in half) but the new param uses percentage to insert the boxes next to a pre-existing box

[cldougl]

ok got it 👍

[cldougl]

@Kully did you get a chance to update this to 2.5.0 and update plotly/version.py as well?

[Kully]

no thank you for reminding me

[Kully]

hey @Kully this looks good, I'm trying out some examples now! Are these changes intended to be backwards compatible?

As far as I know, yes. fill_percent is defaulted to 50, so it behaves the same way before the changes.

There are a couple technically non-backwards compatible things:

• The main container in an init dashboard dashboard.Dashboard() that holds all the boxes has size in %, not px. It's a small technical thing.
• the default height and width of the HTML preview are bigger.

[cldougl]

does this output for the example look correct to you @Kully ? Not sure right/left or above/below is working

[Kully]

Running the exact same code in my branch, I am getting this preview:

My output is correct but yours is not. What could be different about our branches? (you are cloning from the latest commit I assume)

[cldougl]

@Kully running on your branch in python 2.7 I get the desired output but in python 3.6 I get:

We should make this consistent across python versions.

[Kully]

wow, that's interesting -
I'll investigate that (and probably need to add better tests)

[Kully]

@cldougl Now the API auto-sizes very similarly to the dashboard GUI. Ready for I believe a final look.

[cldougl]

@Kully for python 2.7 and 3.6 I am getting this preview from the example:

Can you confirm that is correct? That is not what I was expecting from

my_dboard = dashboard.Dashboard()
my_dboard.insert(box_1)
my_dboard.insert(box_1, 'left', 1)
my_dboard.insert(box_1, 'below', 2)
my_dboard.insert(box_1, 'right', 3)
my_dboard.insert(box_1, 'above', 4, fill_percent=30)

[cldougl]

@Kully can you also add an example of the easiest way to add 4 vertical boxes of the same height?

[Kully]

@Kully can you also add an example of the easiest way to add 4 vertical boxes of the same height?

# 4 equally stacked vectical boxes
import plotly.dashboard_objs as dashboard

box_1 = {
    'type': 'box',
    'boxType': 'plot',
    'fileId': 'username:some#',
    'title': 'box 1'
}

my_dboard = dashboard.Dashboard()
my_dboard.insert(box_1)
my_dboard.insert(box_1, 'below', 1)
my_dboard.insert(box_1, 'below', 1)
my_dboard.insert(box_1, 'below', 3)

[Kully]

Can you confirm that is correct? That is not what I was expecting from

Yes this is correct.

This is perhaps the weakest point in the API currently. box_id numbers (the numbers that appear in the HTML preview and correspond to a box in the layout) are generated on-the-fly. This just means that you need to look at where the numbers are placed after each insertion/deletion of a box to appropriate reference the correct box_id.

Does this make sense?

[Kully]

I am going to edit plotly/dashboard_objs/dashboard_objs.py now so that box_1 is rewritten as box_a to avoid confusion. You suggested a similar variable-renaming to the docs a while ago.

[cldougl]

💃