Ask HN: What Was Wrong with Display PostScript and NeWS?

2 points·tlack··2 comments

My frustration with HTML and Javascript as a mechanism for conveying interactive information grows by the day, so I sometimes study different approaches to building graphical applications.

NeWS and Display PostScript seem way ahead of their time. I never got the chance to use either one. Why did they fade away?

View on HN

I used Display PostScript back in the early days at NeXT. It had a very fundamental performance problem. Since PS was designed as an interpreter and could first be fully processed before rendering a page, DPS had essentially the same design. That meant that you had to parse the entire submitted DPS source before you could start doing any rendering (since there could be dependencies late in the submitted code which would influence earlier code.) So it wasn't a simple Z-ordered geometry - it really was a complex language. So, in practice, it was slow and there was no way to really optimize it without breaking compatibility. We were the first real users and thus the first to realize that we couldn't really compete with either simpler or more sophisticated technology (one faster, the other providing more features.) So it worked for our application as an OS GUI, but it was in many ways a dead-end.

I think they both failed due to high price and overshooting the hardware available at the time.

IMO NeWS had many of the same problems as HTML+JS today, in terms of forcing a frontend/backend split that are written in different programming languages. It's twice as much to learn.

AFAIK DPS isn't that different from today's widely-used Cairo/Skia/CoreGraphics/Direct2D APIs except it had a verbose textual encoding and an extra layer of Turing-complete programming language that you don't need but can allow weird DoSes and such.

Reply on news.ycombinator.com