Description
The steering group has now had the opportunity to take up the proposal to make changes to the inheritance principle (e.g., in #1003).
We discussed these proposed changes in a recent meeting together with the maintainers. Our discussion was intended to assess the feasibility, impact and importance of such changes in moving the standard forward within the 1.0 series, especially as it pertains to the evolution of BIDS derivatives.
Though we realize that these changes could bring clarity to some of the issues brought up in the discussions on the development of some of the derivatives (e.g., bids-standard/bids-bep016#50), we also wanted to assess the impact that these changes would have on already-existing datasets and on the way that existing tools would operate with datasets that implement changes to the principle.
As noted in our meeting and in some of the discussions on GitHub (e.g., Lestropie#5 (comment)), it is our understanding that changes to the inheritance principle can be introduced without breaking backwards compatibility. However, the main risk we foresee in introducing these changes within BIDS 1.x is that existing tools (mainly those reading BIDS-formatted data) would operate with new datasets under faulty assumptions, which would be hard to account for, potentially introducing subtle errors (as described in the aforementioned comment). We recognize the potential value of the proposal in reducing duplication of metadata, but do not consider the metadata duplication a show-stopper for continued development of derivatives under BIDS 1.x.
Therefore after careful consideration of the implications of proposed changes to the inheritance principle we have decided to push these back to a future BIDS 2.0 and to keep using the inheritance principle as-is within BIDS 1.x. We suggest that developments proceed apace with the options that are not in violation of the inheritance principle for now.