Power BI Designer saves a local PBIX file, which can be a file to export data and data model – in other words, it’s a format that contains a complete semantic model. All the applications that today e...
If the template pbit file format was a better designed published textual (then zipped) format; and didn't change ids, inject runtime data into design-time artifacts, and coped with some level of missing referential integrity without reporting the pbit as corrupt, this would enable a lot of additional 3rd party tooling and be a massive improvement to the product.
Coupled with the ability to publish a pbit template to the service directly via API in order to update a report; without having to load and refresh locally into a pbix.
This would enable
- Creating useful APIs to manipulate designs programmatically from a variety of languages (as per this idea)
- Splitting reports into subsets of tabs could be automated (solving tab level permissions idea currently #2)
- Proper version control of source with branch/diff/merge in git or other vcs (my personal highest-priority!)
- Pre-processing for formatting to maintain consistency across reports designs
- Programmatic tab generation - e.g. for monthly tabs or by region
- Switching of data sources from dev to prod environments at deployment so that IT can develop reports with proper segregation of duties
- Analysis of field usage within reports to enable automated optimisation
- Analysis of field usage across reports to see impact analysis of shared data model changes
- And lots of things we haven't even imagined!
I'm a programmer, and have experimented with the files extractable from the pbit; it's already 95% of the way there - it would be so good to finish this properly!
