News for hackers. Stuff that matters.
Re:Electron 4.0.0 by CyberDildonics on Friday December 21, 2018 @11:01AM (#18734101)
The biggest issue by far with every electron based program I've used is actually not the huge memory foot use or the hundreds of megabytes for a simple program, it is the latency of the interface.

Every interaction ends up lagging like molasses, whether it is text input, window resizing, startup time etc.

We have video games that run 3D environments with gigabytes of geometry and textures at 144hz yet people are making simple text editors and interfaces with a dozen simple components (like the ethereum wallet) that run xwindows streamed over a 56.6 modem.

Re:Electron 4.0.0 by jeffbax on Friday December 21, 2018 @11:56AM (#18734574)
Nevermind the small but incredibly obvious stuff like keyboard shortcuts, drag and drop behavior and text selection…
Re:Electron 4.0.0 by spiralganglion on Friday December 21, 2018 @12:26PM (#18734784)
I think this is an excellent criticism of Electron that seems to be buried. Sure, it’d be nice if the RAM usage and disk footprint of electron were smaller. Maybe the creation of some sort of shared single instance of chromium will happen, someday. But even if it does, the browser rendering process is really latent unless you’re extraordinarily careful. It’s very challenging to make sure that your app responds to input within one frame — batching updates with requestAnimationFrame creates a visible delay when dragging and typing. It’s possible to do better, but it’s hard and almost nobody does. And it’s easy to do a lot worse — RAIL isn’t aggressive enough.

(My angle on this — I do high-performance 2d/3d animation work in the browser and in electron, with extensive testing of latency, energy usage, frame rate stability, etc)

Re:Electron 4.0.0 by ahartmetz on Friday December 21, 2018 @12:54PM (#18735034)
I've heard a similar story about an app implemented in HTML5 and node.js (no good JS UI frameworks existed at the time). The fine-tuning took an extraordinary amount of time. I was on a project that redid it in QML in a fraction of the time with a fraction of the headcount and performance was great right away.