Replies: 1 comment 1 reply
-
Hi there! 👋 Great question — and a very valid performance concern to raise. You're right, currently GLSP by default always uses full model updates, as we felt it is simpler for the usual case and in most of the scenarios, the model transfer doesn't actually impact performance a lot. However, there may be scenarios with very large models where a delta-based approach or sending a compressed model makes sense. But this would need to be implemented in a customization implementation at the moment. For those cases it needs to be evaluated what's actually faster: computing the difference, vs compressing the model and sticking with snapshots. Maybe you have a delta available anyway, but that certainly depends on the model source and your model management. But I would only go there once performance becomes an issue and the model transfer has been identified as the main cause. I hope this helps! |
Beta Was this translation helpful? Give feedback.
-
hello,
I am currently developing a tool based on glsp/sprotty, it seems every UpdateModelAction is full update. I am thinking whether along with the diagram become big enough, maybe the serialization/deserialization and network transport will be a performance problem.
I checked the sprotty UpdateModelAction/UpdateModelCommand,it seems sprotty supports delta update. but glsp removed it
And I checked the node server and java server code, both are full update.
My question is why glsp don't support incremental update?
And Is my concern appropriate?
thanks!
Beta Was this translation helpful? Give feedback.
All reactions