Metadata visualization improvement

[garrettmflynn]

fix #54

[CodyCBakerPhD]

OOooo, something else I forgot to requestt - would it be possible to add a 'cancel' button on the frontend instead of having to ctrl+C the server each time? (could be useful in the GUIDE as well to cancel a conversion/inspection/upload operation to the best of our ability

[CodyCBakerPhD]

@garrettmflynn Reminder on conflicts here - let me know when you removed the websocket stuff too and I'll take a look at this

[garrettmflynn]

Good to go! Haven't gotten around to the cancel request yet—but I'll see what I can do.

[garrettmflynn]

OOooo, something else I forgot to requestt - would it be possible to add a 'cancel' button on the frontend instead of having to ctrl+C the server each time? (could be useful in the GUIDE as well to cancel a conversion/inspection/upload operation to the best of our ability

Hmm this is somewhat complicated given that HTTP requests are one-off. I've seen a few hacks online that indicate that you can detect if the client aborts the request while it's occuring—but these are not super clean.

The only other way to do this would be to maintain global references to the running bars / processes and have a /cancel endpoint that can clean these up.

Are these solutions something that you would be interested in?

[CodyCBakerPhD]

There's no way to track the PIDs of the processes and send a pkill or similar signal? I know how that might be done on the Python side (so through an endpoint) if JS doesn't have a way - in that case you'd have a 'cancel' button that would call the endpoints which would trigger the process kill?

[garrettmflynn]

Using that approach, we could notify the browser of related pids using SSE events as the processes are created. Though AFAIK this wouldn't abort the endpoint execution itself.

Would that actually achieve what you want? And I'm assuming you really only care about canceling in the parallel demo?

Otherwise, we could probably execute everything in this endpoint as a thread, then use the hacks referenced before to send some kill command to that—which would have references to everything directly and could shut it down that way.

[CodyCBakerPhD]

OK after digging deeper into cancellation, the only/best 'safe' way to do this would be to have it as a specific feature of the iteration process on the backend

I have an idea for how to implement this in a follow up so I'll kick the bucket down the chain for now

[garrettmflynn]

Done!

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 51.00%. Comparing base (68e1159) to head (281f01a).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #55      +/-   ##
==========================================
+ Coverage   50.00%   51.00%   +1.00%     
==========================================
  Files           6        6              
  Lines         100      100              
==========================================
+ Hits           50       51       +1     
+ Misses         50       49       -1     

Flag	Coverage Δ
unittests	51.00% <100.00%> (+1.00%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files	Coverage Δ
...m_publisher/_demos/_demo_command_line_interface.py	0.00% <ø> (ø)
src/tqdm_publisher/_publisher.py	100.00% <100.00%> (ø)