But in all seriousness:
• It was not too long ago when running a browser on under 1GB of RAM was normal.
• Consumer devices are being produced today have less that 1GB of RAM and mobile chips.
• Browser maintainers are doing great work to ensure they work on these devices.
If I can't get an instant messaging website, a social media website, a web based email client and a text doc running in under 4GB of remaining RAM on a 2016 ultra-book something has gone terribly wrong.

@ada My daily driver is only a few years old: 2 GB of RAM, eMMC memory—basically a lower-end tablet in a laptop form factor. I can get RAM use down to 400-500 MB before opening a browser.

I can barely use sites like Facebook and Google Drive; if I do, I can't have anything else open at the same time. I can't even have a graphical text editor open, I use a console-based one.

Kill…meeee…

@ada Wirth's Law at it's finest
@ada I think you need Emacs 😅
@ada I think part of it is hardware manufacturer's basically need people to buy a new device every couple of years, so they encourage wasteful practices by developers. Performance on lower-end systems should be an explicit part of making any site or program accessible imo
@anarchosaurus try telling a developer that perhaps React/Angular are not the best way to build a website that works well for everyone and you will get your head chewed off.
@ada personally I think SPAs are overdone. I agree with you though. We contracted part of our site to someone else and what we got back was something that used angular for the administrative interface and something completely different for the frontend form users would fill out. All in typescript with what looks like on the fly configuration. Tbh, I still haven't quite figured it out.

@anarchosaurus the web does not need to be that complicated sometimes I feel FE web developers want the complexity of building for native because it allows them to show that they can be smart too. But the end result can often be over-engineered.

On the web it is more important to be robust so a simpler solution would probably be better.

@ada Another thing is bandwidth use.

In uncivilized countries like the US, there are a surprisingly high number of people on dialup, on <1 Mbit ADSL, on slow/high latency/metered satellite connections, on unreliable/slow WISPs, and on metered cellular connections.

But, many developers (of both web sites and even native stuff nowadays) assume you're on a 50+ Mbit always-on connection.

@ada Realistically, I'd argue that every developer targeting either the web or targeting desktop Linux should be forced to test their software on a Raspberry Pi 1 B+, 0, or 0 W, with dialup (or with the ethernet interface throttled to dialup speeds).

If your stuff is usable on that, it'll *fly* on actually good hardware and connections. Battery life on modern hardware will be vastly improved. And, your users will probably get whatever they want done much faster.

@bhtooefr @ada Yes, please this. Because for everyone who has a high end gaming rig, there’s probably 5-10 for who the Pi 0 W would probably be an upgrade from their current setup.

@queerhackerwitch @ada I dunno that it'd be that bad, unless you're counting developing nation users on Nokia 1100s - outside of 3D graphics and video decoding, a RPi 0 will perform similarly to an early Pentium III, I believe.

But if you're right, then I need to revise my estimate downward. I'm not sure what to, though - the RPi 1 B+ is the lowest performance widely available desktop-experience computing device I'm aware of, in 2018.

@bhtooefr
Yes, I feel the 'mobile first' brigade are all patting themselves on the back for reactive websites and ignoring bandwidth, connectivity and load on small processors
@ada
@ada I feel old now, even though I’m 23