live feeding restructure

[adrian-pace]

According to #7 time analysis

Time analysis : On stian logs 627 pads. Everything. (times are in seconds)

parsing: 30.954002618789673 parsing aTextes (only in dev) 0.3829975128173828 operation_builder total 46.490999937057495 operation building time 8.0290367603302 paragraph construction time 32.98596119880676 Operation classification time 0.2220005989074707 Operation context time 4.018999338150024

So we will build a system that wait for new input in the parser. If new input :

• parse it
• update operation construction
• update paragraphs (do smth about the operation per paragraphs that might change
• rebuild context.
• update visualizations (saving them to files)
• update colored text.
• Repeat.

We could then build a simple client that display the plots/metrics when they are updated

[adrian-pace]

@pykcel well. it's starting to work:

the video is of bad quality, but still, it's pretty cool https://drive.google.com/file/d/149OOH5jS7Qf_yBZfEo-T4NEwTdXf_5CV/view?usp=sharing

[lbaligand]

Awesome! Well done 🥇

[houshuang]

Looks awesome. Is this done by hooking directly to the database and reading from it? I'm still thinking about how we can best integrate this in FROG.

[adrian-pace]

This one is done directly by parsing the file. However the methods for hooking up with the database and querying it worked in the non-live version so it should be realtively straight forward to modify them. I already have a draft that I need to test on my linux machine.

A few options are for example storing the data or plots to a file and FROG could load them. We could also setup a tcp connection that send the information to FROG such as metrics or even images.