You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
As a datalab user, it would make more sense for datalab operations to be a graph, rather than an ordered list of operations. This way, users could go off on tangents and then cull those operation branches later. This also would make clear which operations depend on outputs of earlier operations. Basically all operations would start as an offshoot of the initial input images in the data session. As operations pull in the output of previous operations, they will depend on those operations in the graph. The idea is to use something like d3 to visualize the graph of all operations and how they connect to each other, and use that to select them to show their inputs or outputs. This will likely involve some backend work to store more information about operations, maybe storing the dependency graph of operations in the models there.
Modify backend to support operation dependencies
Explore visualizations in the frontend for displaying operation graphs
Explore interaction with those visualizations for things like selecting and deleting operations
Come up with a prototype integrating all this together that can be cleaned up and expanded on in the next sprint.
The text was updated successfully, but these errors were encountered:
As a datalab user, it would make more sense for datalab operations to be a graph, rather than an ordered list of operations. This way, users could go off on tangents and then cull those operation branches later. This also would make clear which operations depend on outputs of earlier operations. Basically all operations would start as an offshoot of the initial input images in the data session. As operations pull in the output of previous operations, they will depend on those operations in the graph. The idea is to use something like d3 to visualize the graph of all operations and how they connect to each other, and use that to select them to show their inputs or outputs. This will likely involve some backend work to store more information about operations, maybe storing the dependency graph of operations in the models there.
The text was updated successfully, but these errors were encountered: