Hey All, Question about invalid analytics model sc...
# gooddata-cloud
s
Hey All, Question about invalid analytics model scenarios. We're seeing a lot of situations where Metrics and Insights "disappear" from the UI when we make changes to the LDM. However, when we investigate or explore with the API, we see the metrics are still there just in an invalid state which causes authoring issues because the author cannot directly fix the issue and we need to resort to using the API to clean up the model. Is there a reason elements no longer show up in the UI when they're invalid? It would be much better if we could at least see them in the UI to either fix or delete.
1
m
Hi Steve, thanks for sending this value feedback into us. I will submit this as product feedback on your behalf so our product team can review this and consider this for future developments. Thank you for sharing your experience with us!
p
🎉 New note created.
s
So you're saying this is expected behavior?
m
Hi Steve, yes this is the expected behaviour. However I would like to double-check this - Would you please be so kind as to tell us what “Errors” you’re seeing? You can find them in your Dev tools
s
The issue is: • Not all metrics and insights are shown in the UI
so in theory no actual error except that we cannot see all insights or metrics
We know they are there because of 2 things:
1. we try to recreate them with the same id and we get errors because they "already exist" 2. when we check the model through the API we find things as being invalid and then we can see them in the api resposne
so the issue is that they need to be shown regardless of if they are valid so we can either a) edit to fix them) or b) delete them so we have a clean model
m
I understand your frustration, but at this time those objects are not shown in the UI. So this is why I have already gone ahead and marked this as product feedback and our product team can review this further
s
Ok, just making sure we're clear on the issue since its technically not an "error". Just issues with authoring.
j
Thanks for the feedback. Btw. You can check impact of model changes in right panel of modeler before you save the model. Many operations are possible without breaking metrics, insights and dashboards. You can for example move attributes to other datasets while keeping their identifiers instead of deleting and recreating with different identifier. Similarly you can map sql datasets to physical tables and vice versa without recreating. Could you tell if your problems are mainly caused by changes of model or rather changes of metric identifiers in metric editor?
s
We've been trying to find a good way to Author LDM and Author Analytics Model. Because authoring the analytics model is tricky without having a parent/child relationship in place so we currently have the LDM in the authoring parent and the analytical model in the child workspace. Which means that when editing the LDM we don't see the impact of the changes on the analytical model. We will probably have to alter how we do LDM authoring since we do want both models to be together. It just is impossible to properly author analytical model items without the parent/child relationship established.
we also have been working/in place the process of validation of both models and the import/export of the authoring environment designs. It works well for analytical model authoring. Does not work well for LDM authoring
j
Thanks for clarification
1
p
🎉 New note created.