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
When designing the Riesgos-frontend, the assumption was that most products would be fairly standardized.
It was thought that the web-app should take a process-output and transform it into a displayable layer by itself.
This is what our layer_marshaller was designed for: accept inputs, detect whether it contains vector- or raster-data, and transform it into a UKIS layer.
By now we have many customizations to our layers, though. There are few layers which don't have deviations from our standard settings.
Examples:
available earthquakes should have a popup appear on hover, not on click
deus outputs need to be rendered in webgl, because they are too large for simple vector-layers
lahar-outlines should be bundled into one single layer with a timeslider
shakyground-outputs need to have their names changed
Our original intention was that a user should only have to understand the interfaces WpsData and Product and their descendants; they should not have to understand Ukis-types. A programmer would only touch the contents of src/app/riesgos/scenarios. But with so many customizations this is getting unrealistic.
I'd now propose that instead of having different Product subclasses which are configured in src/app/riesgos/scenarios, instead we have each product have a method toUkisLayer(): Layer
This allows a programmer greater flexibility in creating a custom Ukis-layer from a riesgos-product. It also avoids having to add many special cases to layer_marshaller. It would force the user to implement that method for every single product, but we're already doing that for many layers anyway. This way that transformation would be more direct and explicit.
The text was updated successfully, but these errors were encountered:
When designing the Riesgos-frontend, the assumption was that most products would be fairly standardized.
It was thought that the web-app should take a process-output and transform it into a displayable layer by itself.
This is what our
layer_marshaller
was designed for: accept inputs, detect whether it contains vector- or raster-data, and transform it into a UKIS layer.By now we have many customizations to our layers, though. There are few layers which don't have deviations from our standard settings.
Examples:
Our original intention was that a user should only have to understand the interfaces
WpsData
andProduct
and their descendants; they should not have to understand Ukis-types. A programmer would only touch the contents ofsrc/app/riesgos/scenarios
. But with so many customizations this is getting unrealistic.I'd now propose that instead of having different
Product
subclasses which are configured insrc/app/riesgos/scenarios
, instead we have each product have a methodtoUkisLayer(): Layer
This allows a programmer greater flexibility in creating a custom Ukis-layer from a riesgos-product. It also avoids having to add many special cases to layer_marshaller. It would force the user to implement that method for every single product, but we're already doing that for many layers anyway. This way that transformation would be more direct and explicit.
The text was updated successfully, but these errors were encountered: