Newton Storage History (2007) | Hacker News Reader

Newton Storage History (2007)

dadhacker.com

78 pointsjsnell11 comments

https://web.archive.org/web/20071208094622/http://www.dadhac...

https://dadhacker-125488.ingress-alpha.ewp.live/newton-stora...

Another one, more technical, on the patching technology we used in the Newton: http://www.dadhacker.com/blog/?p=2233

I loved the whole Newton Soup system. It's a real shame that something like it didn't get further developed. I always wondered what it would look like scaled up into a cluster arrangement.

One of the guys who designed soups (Walter Smith) went to Microsoft and worked on the database-based storage system for Longhorn. It didn't go that well; my take is that the project got really big and politicized -- classic Microsoft -- and he wound up leaving after a while.

I think that systems like ReThinkDB are really close, if not superior. One thing the Newt never had to worry about was concurrency; the scripting environment was single-threaded, and the storage system only ever had one transaction at a time (with some advanced cheating going on for background storage operations and garbage collection). Modern systems solve harder problems, albeit with many orders of magnitude more resources and way better tools.

It should have had, but never got, "contexts" so when your code is changing an object in one place, your other code doesn't see the changes unless you want it to. Ah well, people seem to have made it work anyway. :)

CouchDB is pretty close in principle but does a lot more stuff. For something closer to the original, I think you could do it PostgreSQL quite easily. But then you need a language layer...I think ECMAScript proxies might work. Hmm, I guess somebody should make an npm module!

This is slightly off-topic but has anyone blogged about Palm history as wonderfully as kabdib/Dadhacker/Landon has about Newton and other topics?

Seems like Steve Yegge worked on palm, I haven't read anything from him at this level though, I am not an authority though!

A good enough excuse to troll through his blog posts again at least. ;)

The PalmOS, at least its early iterations, was not nearly so interesting. There was no removable storage or other problems to tackle.

I tended to view Palm programming as an exercise in writing C code that had be written with the intervening 20 years of aggregate experience on good design.

The great thing about Soups/Stores was the sharing: applications could augment the frames with their own data, and everything worked together cooperatively with ubiquitous access to the information.

In the modern era of sandboxing, this is impossible, and services don't compose nearly so well.

The Soup was cool and really ahead of its time. But then I also think that the Newton was slow and a battery hog compared to say a PalmPilot, partly because one had a fancy innovative OS and the other was relatively simple.

Reply on news.ycombinator.com