[ePix100][Correct] Correcting one train for epix100 and storing a list of one pulseId

Description

A few weeks ago a bug was identified in ePix100 correction notebook when correcting a sequence file of 1 train.

This is a fix for the following error when converting an array of one pulseId to a list:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-44-b0dfb1190f77> in <module>
----> 1 list(seq_dc[instrument_src]['data.pulseId'].ndarray().squeeze())

TypeError: iteration over a 0-d array

How Has This Been Tested?

Unit tests and attached the correction result for the run which exposed the bug in the first place

Relevant Documents (optional)

The failed run after correcting it with this MR

Types of changes

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

Checklist:

Reviewers

@duarten @schmidtp

added Testing label

assigned to @ahmedk

changed title from Correcting one train for epix100 and storing a list of one pulseId to [ePix100][Correct] Correcting one train for epix100 and storing a list of one pulseId

changed the description

added 1 commit

• 6e67150f - preserve pulseid dtype

Compare with previous version

changed the description

changed the description

I'm not sure this is the right fix - isn't the .squeeze() the cause for this issue in the first place? In the case of a single train, you're unintentionally squeezing away the train axis. Instead, you could explicitly remove the axis you don't want to have, I suspect it's some (N_TRAINS, 1, x, y) shape?

So the shape is (n_trains, 1)

Of course sorry, we're looking at data.pulseId here. Still the point stands, instead of squeezing (and turning a single train into a scalar), you could simply remove the second axis:

.ndarray()[:, 0]

We intentionally tried to make sure that the dimensionality of data accessor methods in EXtra-data does not change when inspecting a single train.

changed this line in version 3 of the diff

Thank you. Yes, this is differently better. While writing the fix I also felt that I was complicating things more than I should.

added 32 commits

• 6e67150f...6f3f34e7 - 29 commits from branch master
• cdbdc0ef - Correcting one train for epix100 and storing a list of one pulseId
• 03ff42d9 - preserve pulseid dtype
• 75f4e499 - Improve fix by just discarding the axis instead of squeezing

Compare with previous version

LGTM

Thanks @schmidtp for the review!

merged

mentioned in commit 017f4bdc

changed milestone to %3.13.0