Closed Michael-Sun closed 2 months ago
yeah I was thinking of possibly using a proper workflow manager for this. nipype / pydra or bidsnake?
this may not happen fast but it would be useful to have
for now I may make the default behavior to reuse pre-existing files and pass a --force
flag to overwrite previous results
Not sure if you are using it. But splitting the processing by running one action at a time may help a bit as well.
https://bidsmreye.readthedocs.io/en/latest/usage_notes.html#named-arguments
for now I may make the default behavior to reuse pre-existing files and pass a
--force
flag to overwrite previous results
files that already have an output should now be skipped unless the --force argument is passed in the CLI
Is there an existing issue for this?
New feature
Running bidsmreye on a cluster compute gpu node can run into issues if the process gets killed because of memory errors, timeout, or other miscellaneous issues (e.g., administration-level malicious code detection wrongly kills your process). It would be good if bidsmreye can pick up where it left off instead of starting processing all over again.
Unclear documentation
No response