I'm pushing GridWhale, which is a completely server-side platform, but the trick is that we remote the UI to the browser (think X-terminal but at the granularity of UI controls).
The nice thing about this is that your program doesn't have to worry about the front-end/back-end split. You just write as if you had full control over the machine, and the platform takes care of remoting to the browser.
EDIT: 15 minutes after this post, I've already had 20 clicks, and each click spawns a new server-side program. And it's just running on a little 4-core Xeon PC under my desk.
I'll post again when it all falls over!
EDIT2: Not as bad as I feared/secretly hoped. After an hour there are about 100 program instances, but obviously most are idle, so they don't take up too many resources. The compute process is only using 32 MB of working memory, so we have plenty of room for more. My guess is we could easily hit 1,000 program instances. Of course, this kind of traffic is trivial for a standard web server--we're talking something like 50 requests per minute. But you gotta start somewhere!
EDIT3: Spoke too soon! Finally crashed. Although not from traffic but from a bug, so now I have something to debug.
Why is every clickable element implemented as a div? You lose out on a lot of native things, like changing mouse cursor on hover, keyboard navigation for users with special accessibility needs, etc.
I think what he's trying to say, is that's a lot easier to convince someone that one thing is a good idea, rather than about 7 untested, unknown things all at the same time. Conventions are, well, conventional for a reason.
Absolutely true! And that's one of the 1,000 reasons why GridWhale might fail.
There are already lots of people working to simplify web development by fixing one thing at a time. But what if we could start from scratch and build an integrated platform in which all the pieces work together? Would that be better than piecemeal evolution?
Take something like https://Bubble.io. That's an integrated system which has its own custom semantics for UI, storage, etc. But people build apps on it because it's really easy to build apps.
My premise is that there are some people who are looking for an integrated development experience (like Delphi or like Visual Basic) who are not satisfied with either current complex tools or the more limited nocode systems like Bubble.io.
Will there be enough of them to build a business? We'll see.
The nice thing about this is that your program doesn't have to worry about the front-end/back-end split. You just write as if you had full control over the machine, and the platform takes care of remoting to the browser.
Here's the reference manual for it: https://gridwhale.com/program.hexm?id=GCJ5TL7Z&file=GCJ5TL7Z...
[Be gentle--this is implemented as a GridWhale program and everything is still a prototype.]
And here's a post I wrote on the motivation: https://medium.com/@gridwhale/rise-of-the-hyperplatforms-d4a...
EDIT: 15 minutes after this post, I've already had 20 clicks, and each click spawns a new server-side program. And it's just running on a little 4-core Xeon PC under my desk.
I'll post again when it all falls over!
EDIT2: Not as bad as I feared/secretly hoped. After an hour there are about 100 program instances, but obviously most are idle, so they don't take up too many resources. The compute process is only using 32 MB of working memory, so we have plenty of room for more. My guess is we could easily hit 1,000 program instances. Of course, this kind of traffic is trivial for a standard web server--we're talking something like 50 requests per minute. But you gotta start somewhere!
EDIT3: Spoke too soon! Finally crashed. Although not from traffic but from a bug, so now I have something to debug.