module reorganization #53

Activity

1. kushalkolar commented on Dec 6, 2022

   

   kushalkolar

   on Dec 6, 2022

   MemberAuthor

   More actions

   I think we should try and minimize the number of imports that have to be written, try to avoid becoming messy like bokeh. Maybe have the widgets importable at top level since those will probably be used more frequently. Should see what we come up with for the events system as well in #39 and see where that will go.
2. EricThomson commented on Dec 6, 2022

   

   EricThomson

   on Dec 6, 2022

   Contributor

   More actions

   This sounds like a good strategy to me! The one question I have here is where the line is between things that would fall into the Plot and graphics functionality. E.g., again with my usual thought: try to emulate matplotlib where it is strong, but break from it where it is not (and where people don't really use it much anyway).

   import fastplotlib.Plot as plot
   plot = Plot()
   plot.scatter() #does a scatter plot
   plot.plot()  # plots a line (don't make ppl think about what special objects are needed)
   plot.imshow()  # again, just let people do what they already expect
   

   But then for special widgets with a ui component, you need widgets because matplotlib is not great, so you have special things for that, because you don't want to emulate matplotlib.

   Anyway, just thinking out loud here as usual. I may well be wrong, esp as I am not yet a power user of fastplotlib.
3. kushalkolar commented on Dec 6, 2022

   

   kushalkolar

   on Dec 6, 2022

   MemberAuthor

   More actions

   Kind of, this is how the simple plot functionality works:

   from fastplotlib import Plot()
   
   plot = Plot()  # create a plotter
   
   plot.image()  # image plot
   plot.line()  # line plot, put on the same canvas since we're using the same plot instance
   plot.scatter()  # scatter plot, put on top of the previous
   ... etc.

   It's similar to matplotlib but I think a bit more explicity, image, line, scatter are the same exact names as the Graphic classes. That's how it's actually implemented in the Plot class anyways and probably makes it easy for a user to go between the simple Plot interfaces and Graphics if necessary.

   For widgets, currently the image widget works like this:

   from fastplotlib.widgets import ImageWidget
   
   iw = ImageWidget(data=<data>, cmap="plasma")
   
   # if it's a single data array, they can access the `Plot` instance of the imagewidget and add more things
   iw.plot.line(data=<some line data>)
4. kushalkolar

   closed this as completedon Dec 11, 2022
5. EricThomson commented on Dec 16, 2022

   

   EricThomson

   on Dec 16, 2022

   Last edited by EricThomson

   Contributor

   More actions

   Just using with recent changes, it seems there are a lot of imports I'm doing right now, e.g.,:

   from fastplotlib import Plot
   from fastplotlib.layouts import GridPlot
   from fastplotlib.graphics import ImageGraphic
   

   This seems like too much for the simplicity of what I'm trying to do.

   I see the utility of having these layout/graphics things in separate modules, but I wonder if you might want to set up the package (with imports in __init__) so people can just from fastplotlib import Plot, GridPlot, Image and all the things, but also maybe more importantly for the future they can just import fastplotlib as fpl and then just use these different widgets as needed? grid_plot = fpl.GridPlot(...things..).

   FOr people that like to keep things logically ordered, they could still import as above.