I made a computer illiterate friend get one because I was sick of doing tech support for his windows Lappie and he loves it, and has had zero probs with it.. the only time he's called me with a problem with it was when he was trying to some fucking exe to win a free ipod. CHROME os rocks.
There's also in-browser VPNing.
It's like driving a little remote control car to the shops and claiming that you went. You didn't.
I could walk into a library and use a terminal there to type code in, but I'm not doing development on that machine - it's just a dumb terminal.
Now I understand the fact that you can remotely connect somewhere else to do work, but the advent of the microcomputer has meant that you shouldn't need to do that. Power usage for the earth will not be great if everyone runs a device that needs power locally, over a network system that needs powered routers, to a computer remotely that needs power just so you can type code in.
That really doesn't make sense.
I agree that in the context of this (sub)thread, it's a bit odd to claim that chromebooks are decent developer workstations, when what's meant is that they're decent dumb terminals. I'd still like to run my code on my dumb terminal - but I think it makes perfect sense to work like that - have something dumb and wireless for human interaction, and data/processing on a dedicated piece of kit. Be that a low power intel cpu and ssd in a small box behind the tv, or something in a rack somewhere.
And the advent of dynamic libraries has meant we shouldn't need Docker, but, well, that's how it is. It's true that in an ideal world we shouldn't need such a mechanism, but in practice this system works very well for me, and the other ones have been frustrating.
I need a working internet connection anyway to do work: I need everything from software updates to GitHub to IM with coworkers to Google and StackOverflow. And the remote machine shouldn't need any more power than my local one. So my added power consumption is just that of my Chromebook, which is about 10 W.
I admit freely that I am not fulfilling the definition of "development on the Chromebook" in a strict sense, but it is certainly a serious answer for how I develop on my Chromebook.
Have you calculated how much it's costing you?
I ended up scripting up a way to find the cheapest Spot Instance price from Amazon for whatever workload I needed - if I was doing builds I wanted the 8 core machines (c4.2xlarge?). They're ~11c an hour as a Spot Instance, although it's 3x the amount for a Windows instance, and I'd be pushed out of the market if everyone started doing this.
I think I was averaging a couple of dollars a day, so I suppose it comes out in favour of using AWS all the time.
(Also if I had a powerful local laptop, I couldn't tell it to do a build, close the lid, and get on the subway without my backpack catching fire. Nor could I have it provide IPv6 to my apartment, host websites behind CloudFlare, provide shells to my friends, etc.)
If I wanted to put some more effort into it, I could certainly do a good fraction of my work (git clone, vi, IRC and email, etc.) on a micro instance. But even a couple of dollars a day would require me to be pretty careful to come out cheaper.
I think my costs would be a bit higher because Kimsufi/OVH/Hetzner are 300-400ms away from me, so the latency is a bit uncomfortable.
When I was in Canada, a Hetzner box was 200ms away and usable, but for ~$30 a month I only had a dual core box - a bit restrictive when doing large builds.
I might have a go at automating setting up a dev environment on AWS which can be torn down when not in use.
Could you share a couple of lines on how you would set that up? Spin up an upstream-provided basic AMI (eg: RedHat/Ubunut/Debian) - and run a few commands under screen via ssh - and then shut it down?