A developer saves a Power BI report as a Power BI Desktop project (.pbip) and commits it to Git so that the semantic model definition can be diffed and reviewed line by line in pull requests. They want the model's tables, columns, and measures stored as human-readable, text-based definitions rather than an opaque binary. Which format inside the .pbip project provides this?
- AA single packed .pbix binary stored inside the project folder, which Git tracks as one file and compares byte by byte during pull-request review.
- BA compiled .bim resource embedded as base64 text in the report file, which the reviewer decodes locally before they can read the model definition.
- CAn Excel workbook generated from the model that lists every table and measure on separate sheets for reviewers to inspect during the pull request.
- DTabular Model Definition Language (TMDL) files, which express the semantic model objects as text so that diffs and merges show meaningful per-object changes. Correct
Why A is wrong: A .pbix is a binary container, so Git can only treat it as one opaque blob, which defeats the readable per-object diffing the developer wants from the project.
Why B is wrong: Embedding a base64 blob is still effectively binary and must be decoded to read, so it does not deliver the plain-text, directly reviewable model the project requires.
Why C is wrong: A workbook is a separate export, not the project's source format, so it is not version-controlled with the model and cannot drive line-by-line diffs in Git.
Why D is correct: TMDL serialises the semantic model into folder-based text files per object, giving the readable, diffable definitions the .pbip project needs for review in Git.