A fresh install of Signal takes up 410MB, blowing both Firefox and Chromium out of the water
A fresh install of Signal takes up 410MB, blowing both Firefox and Chromium out of the water
... and I can't even continue the chat from my phone.
You're viewing a single thread.
Like I know native apps are always better, but why doesn't electron ship an installable runtime so we don't have to have a shitload of inert chromium installs on one machine?
May be, but I don't think apps use it. Afaik Teams, Discord and such are all epectron apps, yet they have not much in term of dependencies and large install sizes, so they must ship with their own versions.
You don't understand. This way if some app crashes it will not cause others to crash too.
This is how google introduced the "multiprocess architecture" of Chrome.
You can still have separate processes and everything else with a shared runtime, you just save having all this wasted storage with every application bringing its own bundled runtime.
.net or Java applications work in a similar way, one Java app crashing won't take out another just because they're sharing the same runtime
I'd rather not have frameworks based on web browsers. Programming is not that difficult.
For most uses of electron I'd agree, but if some engineers are going to use it anyway, I'd prefer the approach I've described.
Programming is not that difficult.
Learning how to do something in a new language and framework isn't that tough, I agree, but no one is going to become an expert in something overnight. I don't reckon many desktop native engineers are choosing electron unless they actually need it, so if you imagine the case of an expert web engineer building a desktop UI, they're going to do a much better job with their main skillset than something they have just learned.
but no one is going to become an expert in something overnight
It's not like they need to become experts. But also that's actually possible (at least the effects of that), especially with all the AI around.
It's not like they need to become experts
I mean if they would produce a better UI by using their expertise, how would not becoming an expert in the new thing be better? The reality is that the people paying the engineer are going to want the better UX over the benefits of not using electron in most cases.
But also that's actually possible
Respectfully, no it's not, not with software engineering unless you're talking about learning a simple library or something.
If someone can genuinely master something in a day it wasn't much of a skill to begin with.
I've been in this industry for about 20 years now, I would find it very hard to believe an engineer who says they've gone from no knowledge to expert in a new framework/language in any short period of time. I would either assume they're trying to pull a fast one or more charitably just in the "naively confident" phase of learning:
especially with all the AI around.
AI can assist you if you more-or-less know what you're doing, but a novice replacing proper learning with ChatGPT pairing is going to write some shitty code. I use AI in my role semi-regularly, and in my experience, no model has consistently produced me anything (non-boilerplate) longer than a couple of lines that didn't need some kind of refactor for it to actually be up to our code quality standards. Sometimes you see them spit out some ancient way of doing things that have been outright replaced by a more modern approach, if you don't have the experience, you'll not know any better.
I mean if they would produce a better UI by using their expertise, how would not becoming an expert in the new thing be better?
I failed to understand the meaning of this sentence. It doesn't make sense to me. Producing a better ui is not even on the table when we are talking ui frameworks and native programming - you use what's available, and if you are a graphics designer then maybe you should've sticked to that instead. Becoming expert in native ui is super cool but I wouldn't expect such miracles from everyone. Just producing a valid low level code is enough to meet my standards of performance. That's because those standards were heavily affected by web frameworks existence.
The reality is that the people paying the engineer are going to want the better UX
And I hoped it would be customers who would pay for a software or a service who would send valid feedback.
AI can assist you if you more-or-less know what you're doing
Assuming web devs creating apps don't know what they're doing?
but a novice replacing proper learning with ChatGPT pairing is going to write some shitty code.
Chances are that code would be much more optimized than anything electron/CEF wrapped.
to actually be up to our code quality standards
Quality standards are great. But seeing companies shipping fixes to simple CSS issues that were breaking some of main app functions made me realize most of them don't care about quality standards. If that's how it is and if there will still be a lot of broken stuff across app updates - might as well just go all the way to proper low level languages.
Well obviously it is, or we wouldn't have electron in the first place.
We have it only because some devs are lazy.