alt Hacker NewsI was going to say... The size of the JS only matters the first time you download it unless there's a lot of tiny files instead of a bundle or two. What the article is complaining about doesn't seem like it's root cause of the slowness.
When it comes to JS optimization in the browser there's usually a few great big smoking guns:
1. Tons of tiny files: Bundle them! Big bundle > zillions of lazy-loaded files.
2. Lots of AJAX requests: We have WebSockets for a reason!
3. Race conditions: Fix your bugs :shrug:
4. Too many JS-driven animations: Use CSS or JS that just manipulates CSS.
Modern mobile network connections have plenty of bandwidth to deliver great big files/streams but they're still super slow when it comes to round-trip times. Knowing this, it makes perfect sense that Nextcloud would be slow AF on mobile networks because it follows the REST philosophy.
My controversial take: GIVE REST A REST already! WebSockets are vastly superior and they've been around for FIFTEEN YEARS now. Do I understand why they're so much lower latency than REST calls on mobile networks? Not really: In theory, it's still a round-trip but for some reason an open connection can pass data through an order of magnitude (or more) lower latency on something like a 5G connection.
Replies
>Do I understand why they're so much lower latency than REST calls on mobile networks? Not really: In theory, it's still a round-trip but for some reason an open connection can pass data through an order of magnitude (or more) lower latency on something like a 5G connection.
It's because a TLS handshake takes more than one roundtrip to complete. Keeping the connection open means the handshake needs to be done only once, instead of over and over again.
I've never seen anybody recommend WebSockets instead of REST. I take it this isn't a widely recommended solution? Do you mean specifically for mobile clients only?
How do you feel about SSE then?
15MB of JavaScript is 15MB of code that your browser is trying to execute. It’s the same principle as “compiling a million lines of code takes a lot longer than compiling a thousand lines”.