Draft: [LPD][JUNGFRAU][CORRECT] Avoid reading INDEX/timestamp for dataFormatVersion 0.5

This bug was identified while correcting old LPD run from proposal 900113.

Description

Avoid reading INDEX/timestamp for dataFormatVersion = '0.5'.

This bug is expected only for old RAW data.

How Has This Been Tested?

Relevant Documents (optional)

Types of changes

• Bug fix (non-breaking change which fixes an issue)

Checklist:

Reviewers

@schmidtp

marked this merge request as draft

https://git.xfel.eu/detectors/pycalibration/-/blob/master/src/cal_tools/files.py#L305

md_group.create_dataset('dataFormatVersion', shape=(1,), data=b'1.2')

What shall the dataFormatVersion be in that case? Still 1.2? I guess this is a different dataFormatVersion compared to the RAW data one? As I see this was hard coded

Yes, the API is intended to write data in exactly one format version, preferably the newest.

changed the description

I understand the problem, but I'm not happy with the solution. Notebooks should not care for such things as format versions, it re-introduces the mess of manual file writing we tried to get rid of with EXtra-data. This particular code section is unfortunately itself not a proud example of doing this.

Given this exact same problem will happen in most notebooks that do not write data en-bloc, we should use the DataFile API to hide it, it's what it was meant to be. I will mess around with a different approach.

mentioned in merge request !752 (merged)

Closing this MR, as there is a more optimal fix available to review at https://git.xfel.eu/detectors/pycalibration/-/merge_requests/752/diffs

closed