Client performance when moving panels about phosphor HOT 8 CLOSED

It really depends on the contents in the panels. The act of resizing the panels themselves is < 1ms on my machine. If you are reflowing lots of stuff within the panels, the cost can add up.

Try running this example and seeing how it affects your CPU usage. It's negligible on my machine.

from phosphor.

Also, the chrome profiling dev tools can help you pin down the costs during resize with nice flame charts.

from phosphor.

I'm totally out of my element here when it comes to browser performance, but one thought would be to throttle resize events to occur no more often than every 20ms or something.

from phosphor.

Here's a snapshot of the timeline during panel resizing on my machine. You can see that a vast majority of the time is spent idle, and only a very small part is spent in scripting.

from phosphor.

Splitter resizes are already throttled on animation frames (which let's the browser pick the optimal refresh). My hunch here is that you have content in your panels which takes a long time for the browser to relayout due to the resize. What are you displaying in your panels?

from phosphor.

It first arose when when dealing with Bokeh plots (which would make sense as expensive) but I can reproduce with a notebook and a terminal

from phosphor.

This is almost certainly inline reflows due to resizing the notebook, and not an inherent performance problem with Phosphor. Can you please open an issue on JLab so we can track it down there?

from phosphor.

Way ahead of you: jupyterlab/jupyterlab#958 :)

from phosphor.