UCSD Pascal pioneer Ken Bowles has died

Re:UCSD Pascal pioneer Ken Bowles has died by WalterBright on Sunday October 07, 2018 @03:35PM (#18162244)

UCSD Pascal was the first "Java" in that Pascal was compiled to machine-independent P-Code that was then interpreted. (I used it in the 1980s.) The main problem it had was it was slow. In the 90's, even with the CPU advances, interpreted Java was still too slow. It took Symantec's Java JIT compiler to really make Java work.

I.e. what UCSD-Pascal lacked was a JIT compiler. Even a simple one would have made a huge difference.

UCSD-Pascal was the right idea, it was just before its time. I think Java owes a lot to Ken Bowles.

[ Parent | on HN ]

Re:UCSD Pascal pioneer Ken Bowles has died by everybodyknows on Sunday October 07, 2018 @04:06PM (#18162368)

By '80 or so it was possible to link to assembler, and the recipe was to code inner loops in assembler. With a Digital Equipment LSI-11 CPU and its simple, highly regular machine code, this was workable for a time.

[ Parent | on HN ]

Re:UCSD Pascal pioneer Ken Bowles has died by protomyth on Sunday October 07, 2018 @05:47PM (#18162798)

LSI-11 http://www.cpushack.com/2017/11/22/cpu-of-the-day-dec-lsi-11...

[ Parent | on HN ]

Re:UCSD Pascal pioneer Ken Bowles has died by mark-r on Sunday October 07, 2018 @10:40PM (#18163793)

P-code was an absolute innovation. Its time in the limelight was short lived, as Turbo Pascal soon came along and compiled directly to Z80 and later x86 - it killed UCSD Pascal in performance.

I borrowed the idea around 1980 to develop a language for doing data entry and verification. The trick was to make the instructions really high level, so the overhead of interpretation was a small part of the total.

[ Parent | on HN ]

Re:UCSD Pascal pioneer Ken Bowles has died by WalterBright on Monday October 08, 2018 @12:36AM (#18164166)

The JIT does for interpreters what the Merlin engine did for the P-51. Originally, it had the Allison engine, and was sluggish and performed poorly. The airframe lit up when fitted with the Merlin engine.

[ Parent | on HN ]

Re:UCSD Pascal pioneer Ken Bowles has died by pjmlp on Monday October 08, 2018 @03:19AM (#18164867)

It was possible to AOT compile to native code with UCSD Pascal, Western Digital and Corvus Systems supported some form of it.

I thought it was Strongtalk/SELF JIT compiler that made Java work.

[ Parent | on HN ]

Re:UCSD Pascal pioneer Ken Bowles has died by tomjakubowski on Sunday October 07, 2018 @05:53PM (#18162827)

TIL Symantec made the first Java JIT! Did HotSpot use any code from their implementation or was it more of an inspiration?

[ Parent | on HN ]

Re:UCSD Pascal pioneer Ken Bowles has died by WalterBright on Sunday October 07, 2018 @07:33PM (#18163197)

My colleague Steve Russell was the implementer of the JIT. Steve was a true wizard with x86 assembler code. You can see his work here:

https://github.com/DigitalMars/optlink

I don't think HotSpot used any code from it, but I can't be positive about that. But Symantec's JIT was so obviously a huge improvement, every vendor had to have it or be left in the dust.

[ Parent | on HN ]

Re:UCSD Pascal pioneer Ken Bowles has died by mwcampbell on Sunday October 07, 2018 @11:11PM (#18163898)

> My colleague Steve Russell

Is that the same Steve Russell that hacked Spacewar and implemented the first Lisp?

[ Parent | on HN ]

Re:UCSD Pascal pioneer Ken Bowles has died by WalterBright on Monday October 08, 2018 @12:29AM (#18164134)

He's way too young for that :-)

[ Parent | on HN ]

Re:UCSD Pascal pioneer Ken Bowles has died by flyinghamster on Sunday October 07, 2018 @10:53PM (#18163845)

Digital's BASIC-PLUS on RSTS/E also compiled to an intermediate code that was interpreted. However, unlike P-code (or Infocom Z-code for that matter), it wasn't portable to other systems.

[ Parent | on HN ]