Forum Discussion
Developer mode & git for reports with live connection
No, when you open in Desktop and its a byPath it will also load the model from disk and let you edit it if you like. This allows Desktop developers to work in complete isolation without service dependency.
I understand, that if you are working with a very large model this could be inconvenient, because you are forced to at least once refresh the model locally. As a workaround, you can:
- Edit the definition.pbir and switch to a "byConnection" (it's on the backlog to provide a UI feature to facilitate this swap)
- Move the semantic model or report to a different workspace
I'm also curious to understand what, in your opinion, should be the right behavior?
Thank you for your explanations, RuiRomanoMS
For me, the live connection must remain active without any intervention from the developer.
The Report-"Developer" could be anyone without any knowledge of data modeling, editing JSON files, etc.
Having a report with a live connection switch to a local data model makes no sense. Then, I must edit a JSON file (or change a setting) to switch to the expected behavior.
Why?
Any other opinions on this, guys?
Thanks for the feedback.
A few follow ups:
- Why those report developers need to go to Desktop and dont edit/develop the reports in the service?- Are those report developers confortable enough with the git concepts: cloning, merge, commit, push? If not, the download PBIX & upload/publish may be a better and simpler alternative.
- If we always export the report with a connection to the model in the workspace, let's say you have 1 model and 2 reports, what would be your expectation if you sync that git branch to a new empty workpace? It creates the model and the 2 reports connected to the previous workspace?
>> Having a report with a live connection switch to a local data model makes no sense
The report does not switch to a local mode. Contrary to PBIX, in PBIP the report is completely separated from the model (they are two distinct folders). When you publish back to the service, using Git, the relative/local connection is resolved to a service connection.
>> Then, I must edit a JSON file (or change a setting) to switch to the expected behavior.
I agree, it should not require a JSON file edit, in the future we want to have a simple UI for this.
So, why do we need Git and version control when part of the solution is outside of Git?
It is easy to explain, even to non-technical-"developers", the concepts of Version control.
So, yes. They should work with Git and sync their work through it with the Workspace.