Fmriprep debug flag

I’m considering if opening a documentation issue or enhancement is needed for the following:
In order to understand the differences when using the “debug” flag (currently renamed to- “sloppy”), one needs to review the whole code of fmriprep…
Could the differences between the standard and the “debug” execution be summarized in the documentation?

The two following things should be true:

  • No workflows connections are changed by --sloppy
  • The parameters of any task may be changed by --sloppy to exchange quality for speed

I think the only place it has any effect right now is on some registration steps, but we don’t make any guarantees that it won’t spread, and you should not use it. If it has any effect on your workflow, it will not be good.

Feel free to open an issue/PR with an improvement to the documentation. Part of the reason it’s under-documented is because it’s entirely intended for internal use. We changed it from --debug because people expected it to provide debugging information (like --verbose), when in fact it means to make speed/quality trade-offs for testing on continuous integration (CI) services.

1 Like