High cpu usage about firefox-better-history

[Ladergorn]

I found that CPU usage by Firefox was about 40% whether in day or month view and this remained constant over several minutes.
When I closed Better History, FF CPU was less than 3%.
If you look at the reviews, others seem to have this problem too.

[bricakeld]

So the CPU usage on mine only increases drastically for about 30seconds then it goes back to normal, and this only happens on the month view. Does your CPU usage increase even on day view?

[argumemnon]

I too experience high CPU usage from Firefox whenever this extension is open. Firefox jumps up to around 200% CPU utilization and stays there as long as the tab is open.

info:
better history version 1.1.0
Firefox 91.0.2
macOS 11.3.1
MacBookPro17,1 (M1)

[bengb]

I get this high CPU problem too, with:
Firefox 93.0
MacOS 11.6 (2018 Intel Macbook Pro)

The peaks at the beginning and end of the graph are where the Better History page is open, and the drop in the middle back to normal levels is when BH is closed.

about:performance shows:

This is on Day view. It doesn't seem to make any difference with Week or Month views....all have the high CPU problem.

It's a great and useful add-on, so hopefully this can be fixed.

[brendonjkding]

@Vincent-P I wonder if the capitalized filename of https://github.com/Vincent-P/firefox-better-history/blob/master/extension/Globe.svg would cause this high cpu usage.
This high cpu usage disappered when I enabled dark mode, which makes me further believe that the capitalized filename is the culprit.

[Vincent-P]

Thank you for taking the time to profile the extension @brendonjkding !
Can you confirm that the issue is indeed related to the name of the svg by cloning the repo, renaming the svg and see if the CPU utilization is better?
If you don't have time to do that I will try later this week.

[brendonjkding]

I tried and can confirmed it now! PR #34 should fix this.

Before renaming the CPU usage would rise to 100% and stay on it until the page was closed and after renaming the CPU usage would go back to 1% around in several seconds on Day view.

I hadn't confirm it at first because when I load the extension manually, there is no high CPU utilization at all, either the one from the repo or the one extracted from .xpi file. I guess this is because the file can be correctly loaded when unzipped, by virtue of the case-insensitive file system of macOS, but can't be loaded when zipped into a .xpi file. And I just realized I can simply zip them to see if there is any difference. And the result is significant as mentioned above.