[Suggestion] sourceLabels as 1D array

[HanBnrd]

Currently, probe.sourceLabels must either be of shape 2D <n_sources> x 1 or 2D <n_sources> x <n_wavelengths>.

Although the specs say that sourceLabels must be a string 2D array, could we allow a string 1D array too, because in practice most SNIRF files do have this as a 1D array. For example:

• https://github.com/fNIRS/snirf-samples
• https://openneuro.org/datasets/ds005964/versions/1.0.0

[samuelpowell]

Do we know why this constraint was originally applied? For example, is the idea that this is always n_wavelengths in the second dimension?

[HanBnrd]

Do we know why this constraint was originally applied? For example, is the idea that this is always n_wavelengths in the second dimension?

I don't think so because I assume x 1 would be the most commonly used and mean that sources from both wavelengths have the same name

[samuelpowell]

@sstucker you don't happen to recall if there was a specific reason for this approach?

[Edouard2laire]

Is there any data available storing a 2D array for sourceLabels ? The previous validator in pysnirf2 is validating files using 1D arrays for this field. Maybe we can switch to only using 1D arrays, as we do for the detectorLabels.

For the wavelength information, it is handled by probe.wavelengths so it doesn't really need to be present in the sourceLabels.

[HanBnrd]

The validator seems to not follow exactly the specs on that one (you can find # Override sourceLabels validation, can be 1D or 2D in pysnirf2).