PyQt5 end of life is coming in mid-2023. At some point, it's time to move to PyQt6. (It's almost 2 years since it was released, but still no conda packages?) Needs updates to requirements, import statements and possibly minor code changes too.

Alternatively, investigate if it's feasible to move to PySide2, which is the official Qt for Python binding. Last time I checked I didn't want to do it, since Qt Group was being tricky with licensing or something.

(It's also possible to support both PySide2 and PyQt by conditional import statements and some other minor trickery, but this is probably not necessary)

Even if the progress bar itself does not obscure the feet, the player window adds a semiopaque area around the progress bar that degrades the video quality in the feet area. The problem doesn't happen for the overlay videos as they have a black frame around the actual video.
One straightforward solution is just to add a black stripe below each non-overlay video.

Currently postprocessing pipelines sometimes fail, for all or some of the videos, without notifying the user. Add the checks that the expected number of videos has been created.

Add session-level notes to the web and pdf reports
Change trial level notes:

• E1, running and E1 running should be treated as E1 for the purpose of inclusion into web or pdf report. running is a comment here and should be displayed somewhere in the web/pdf report.
• If there are trials toe standing and toe standing continued, both should be included into the web report. This should work, at least, for the toe standing and unipedal ... trials (Can this be easily extended to all trials?).

Hello Jussi.
When we use your scripts to process a trial it cuts the trial before we have foot off identified on the forceplate. It processes a full gait cycle, but not foot off on the forceplate, which we need for the kinetics. It is only on the left side we have a problem. Foot length is added in subject measurements. Can we somehow make the region of interest longer ..?

@andreyzhd it would be good to figure out how the events with invalid context happened, i.e. is it possible to create Nexus events that gaitutils won't accept?

The event reading code is in nexus._get_metadata() and c3d._get_metadata(), for Nexus and c3d files respectively.

When creating a comparison web report from two sessions:
The order of the videos in the report depends on the order in which the sessions were chosen, but ...
The order in which line styles (solid, dashed) are assigned to the sessions does not.

That is, reverting the order in which the sessions are chosen results in videos being swapped, but the plots remain exactly the same.
The desired behavior is for the plots to change too.