We have found that having to manually add a new dataset to the index.aml file quite annoying and our engineers regularly forget this. It would be great for this to be done automatically, especially now with the git integration coming out. A user can create a dataset and when pushed to production it appears for everyone else, but whilst not they can still test it at the bottom of the AML pane.
Removing this manual step just makes development of AML a lot easier.
I do agree that the process of manually adding dataset to index.aml could cause friction at some points.
We’re working on the new version of our AML and let’s say if we deprecate the index.aml file, when clicking deploy to production, all the datasets within your project will be published to your end-users in Reporting. By doing that, you don’t need to remember to add your dataset to index.aml anymore.
What do you think of this? Is there a use case that makes index.aml file useful to your workflow?
I think that is fine. From our usage so far, I do not see the need to not add a dataset to the index.aml file especially now with the git integration. You can work on a dataset, test it out in the explore pane at the bottom and then checkin when ready, allowing people to start accessing it.
At the moment the index.aml file is just a step people forget to do and wonder why they can’t see their dataset to use it.
I would love to announce that with the recent release of AML 2.0, adding dataset to index.aml is no longer required.
For now, when you click Deploy to Production, all the Datasets in your project will be presented in the Reporting layer and ready to be consumed by your end-users.