Gatsby before/after

[KyleAMathews]

Couldn't really see any difference. There's a fair bit of variation on timing but the "long frame" time was around ~180ms for both with or without this plugin.

[TheLarkInn]

Interesting! It is supposed to help solve parse time. @nolanlawson can comment more on it.

[TheLarkInn]

I need to setup performance.mark() for debug mode for this plugin however I'm not sure what exactly to measure etc.

[nolanlawson]

You can use performance.mark() and .measure() if you want, but these are not supported in Safari.

I’d recommend just using:

var start = performance.now();
// code goes DIRECTLY here
var end = performance.now();
reportSomehow(end - start);

Also keep in mind that you would need to add a random query param to the end of the script URL if you want to run it multiple times and take the median/average, because otherwise the parsed/JITed result will be cached.

– Nolan

[nolanlawson]

Also FWIW this doesn't just affect V8; it also affects Chakra and SpiderMonkey. See: http://github.com/nolanlawson/optimize-js

[TheLarkInn]

My question is, where and what are you measuring. Where should the start measure go and where should the stop mark go. They answer will help me understand what the hell to look for.

[nolanlawson]

Whatever code you're trying to optimize, that's the code you should wrap in performance.now() measures. You can see how I modified the benchmark library files to see how to do this.

[TheLarkInn]

Ah thats true. I think we discovered the other browsers engaging in this sort of heuristic after I created the plugin. Ugh renaming things.

[addyosmani]

⏩ LazyParseWebpackPlugin

[TheLarkInn]

Lol yes I'm probably going to just change it to that