After presenting notebooks with a thesis committee member today I realized that it would be useful to serialize plots to disk so I that I can quickly bring them back up when a Flatiron cluster node decides to end my session 😆 .
GSP is a WIP but I'm thinking of something much simpler, at least for now.
- Each
Graphic implements to_dict() and classmethod from_dict() that stores all the buffers necessary to create the appropriate geometry and material for that Graphic, this can probably be implemented via methods on corresponding GraphicFeatures.
- Apart from serializing Graphics, I think we just need to store the state of each
PlotArea, i.e. cameras, controllers, canvas size.
GridPlot holds info for the gridplot configuration and all subplots- For
ImageWidget we would need to store the entire 2D or 3D movie, so maybe ultimately store things to zarr instead of hdf5. 🤔
- classmethods on
Plot, GridPlot and ImageWidget?
- Somewhere in the output file store the type of plot that needs to be constructed and the fpl version it was created in. Have a helper function
fastplotlib.load_viz() that auto create the right type of plot
- No animation function support for now
I'll play around with this and note that it's very experimental.
After presenting notebooks with a thesis committee member today I realized that it would be useful to serialize plots to disk so I that I can quickly bring them back up when a Flatiron cluster node decides to end my session 😆 .
GSP is a WIP but I'm thinking of something much simpler, at least for now.
Graphicimplementsto_dict()and classmethodfrom_dict()that stores all the buffers necessary to create the appropriate geometry and material for thatGraphic, this can probably be implemented via methods on correspondingGraphicFeatures.PlotArea, i.e. cameras, controllers, canvas size.GridPlotholds info for the gridplot configuration and all subplotsImageWidgetwe would need to store the entire 2D or 3D movie, so maybe ultimately store things to zarr instead of hdf5. 🤔Plot,GridPlotandImageWidget?fastplotlib.load_viz()that auto create the right type of plotI'll play around with this and note that it's very experimental.