Nothing is wrong with computers. Something is wrong with developers.
“You WILL accept our defined use cases for you. We aren’t interested in writing robust software. We’re interested in writing it badly in 2 days so we can spend the rest of the money on marketing.”
At a small enough company they’re often the same thing.
Also, on mobile, developers who make good, reliable, robust software are discouraged from making such things. There’s a reason there are so few pieces of “finished” software on mobile. Because you’ll invest months of your life into making an incredibly useful and functional tool, and a year later, the new version of the mobile OS will come out, and it’ll be “In the new version we’ve decided half of the operating system calls your software depends on are insecure, and three of the permissions that are necessary for your app to work no longer exist. Have fun rewriting the entire thing”, after which the developer very reasonably says to hell with it, and goes back to writing software for an ecosystem that doesn’t break every single user space application on a regular basis.
So yes, just by working in the mobile space, you’re already accepting trade-offs in making robust software by the very nature of the ecosystem.
That’s a different mechanism. A “page” is just a fixed-size portion of memory, e.g. 4 KiB, which is a convenient size for your OS to do its whole memory management with. And then there’s many things the OS does with such pages.
Page caching keeps files that processes loaded from your hard drive in RAM, after the process doesn’t need it anymore.
What you’re referring to is kind of the opposite. The OS allows processes to reserve more memory than there is physically available. This is called “virtual memory”.
When processes do that, then some of those portions of memory pages get put onto your hard drive, and only get put back into RAM (replacing something else) when the process actually accesses those pages.
The fundamental result isn’t much different, there’s 30 GB of populated memory being sent to disk because these apps all over allocate and the os insists on sending it to disk even with 40% of real ram free.
What they’re saying is that because of this caching, unused RAM is essentially just not a thing. Any process using lots of RAM will slow down everything else on your PC.
200 MB modern application built on top of Chrome can’t handle a few files.
Emacs from the 70/80s can handle a thousand files. Something is wrong with computers.
Nothing is wrong with computers. Something is wrong with developers.
“You WILL accept our defined use cases for you. We aren’t interested in writing robust software. We’re interested in writing it badly in 2 days so we can spend the rest of the money on marketing.”
I’m sure most developers would prefer spending the necessary time to write something good. The problem is perverse incentives in the corporate model.
Sure, for normal software on a real platform. But in mobile it’s often small startups, which means this is explicitly what they WANT.
Again, the developers, or those in charge?
At a small enough company they’re often the same thing.
Also, on mobile, developers who make good, reliable, robust software are discouraged from making such things. There’s a reason there are so few pieces of “finished” software on mobile. Because you’ll invest months of your life into making an incredibly useful and functional tool, and a year later, the new version of the mobile OS will come out, and it’ll be “In the new version we’ve decided half of the operating system calls your software depends on are insecure, and three of the permissions that are necessary for your app to work no longer exist. Have fun rewriting the entire thing”, after which the developer very reasonably says to hell with it, and goes back to writing software for an ecosystem that doesn’t break every single user space application on a regular basis.
So yes, just by working in the mobile space, you’re already accepting trade-offs in making robust software by the very nature of the ecosystem.
This. I also wonder when people will realize that having more RAM doesn’t mean having to use it all (electron, looking at you)
I think most operating systems today are designed to make opportunistic use of available RAM but also fuck electron
Yes, it’s called page cache and you won’t find a mainstream operating system without it.
I regularly get to a 30gb page file on my desktop though…it has sufficient space but still
That’s a different mechanism. A “page” is just a fixed-size portion of memory, e.g. 4 KiB, which is a convenient size for your OS to do its whole memory management with. And then there’s many things the OS does with such pages.
Page caching keeps files that processes loaded from your hard drive in RAM, after the process doesn’t need it anymore.
What you’re referring to is kind of the opposite. The OS allows processes to reserve more memory than there is physically available. This is called “virtual memory”.
When processes do that, then some of those
portions of memorypages get put onto your hard drive, and only get put back into RAM (replacing something else) when the process actually accesses those pages.The fundamental result isn’t much different, there’s 30 GB of populated memory being sent to disk because these apps all over allocate and the os insists on sending it to disk even with 40% of real ram free.
Caching and Electron are two vastly different things.
What they’re saying is that because of this caching, unused RAM is essentially just not a thing. Any process using lots of RAM will slow down everything else on your PC.
I was mainly just making a playful remark about electron being stinky