# Technical Monitoring (minutes) - 2023 April 24

:::info
**Date:** Apr 25, 2023 17:00(UTC)

**Agenda**: [link](https://hackmd.io/@NiPreps-technical-monitoring/r1iXj0zQh/edit)

**Participants:**
- Elodie (ES)
- Chris (CM)
- Celine (CP)
- Mathias (MG)
- Eilidh (EM)
- Oscar (OE)

**Minutes by:** ES
:::
### Quick recap of previous meeting (Apr 11)
The following topics were discussed during the previous meeting:

* SDCflow issue + resampling problem. Two releases of SCDflow in the 2.2 series were made. These releases are fixups. Now SDCflow can handle all orientation issues.
* SynthStrip + namespaces
* SynthStrip in the context of Nibabies

### NiPreps Python namespace
Last week an empty package was pushed to PiPy. This allows us to load packages as nipreps.package-name. Currently, a wrapper of SynthStrip is available as nipreps.synthStrip.
OE proposes to replace the interfaces and workflows in Niworkflows with nipreps.package-name.interface and nipreps.package-name.workflow.
Switching to this solution may be premature because of the work it requires to build the namespace and adapt all the dependencies. However, we can keep this idea for later when it becomes essential and less challenging to implement.

### Provenance tracking
Proposition: Decorate every command line with Datalad run to track the provenance.
Since Datalad is designed to track changes in datasets, each directory must be a Datalad dataset. Implementing this implies restructuring all Nipreps directories.
Something similar is currently used in openneuro (see [https://github.com/openneuroderivatives/ds000001-fmriprep](https://github.com/openneuroderivatives/ds000001-fmriprep) and [https://github.com/poldracklab/tacc-openneuro](https://github.com/poldracklab/tacc-openneuro)).

### fMRIprep-rodents update
* Docker overhaul:
One of the Docker images of fMRIprep-rodents needed an overhaul. The docker image doesn't build locally. fMRIprep and NiBabies started to use wheel packages (see [PR](https://github.com/nipreps/fmriprep-rodents/pull/54)). Is this transition necessary for fMRIprep-rodents?
Since the wheels solution was implemented to avoid a hack to pass the version to Docker, it would be better to use wheels to build the Docker image of fMRIprep-rodents.
* sdcflows integration:
Topup does not work well with unscaled images. EM implemented a workaround to use Topup for images with resolutions smaller than 1 mm. Since fls does not allow resolution below 1mm, it may be useful to generalize the workaround to SDCflow.