home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!olivea!hal.com!bang.hal.COM!not-for-mail
- From: landman@hal.COM (Howard Landman)
- Newsgroups: comp.lang.forth
- Subject: Re: Forth Standard Debate
- Message-ID: <1iknfpINN9jh@bang.hal.COM>
- Date: 8 Jan 93 20:16:57 GMT
- References: <1i7flnINN9or@life.ai.mit.edu> <1993Jan4.234621.12186@news.uakron.edu>
- Organization: HaL Computer Systems, Inc.
- Lines: 45
- NNTP-Posting-Host: bang.hal.com
-
- In article <1993Jan4.234621.12186@news.uakron.edu> r3btg@vax1.cc.uakron.edu (Ben T Galehouse) writes:
- >on 16-bit machines and on 32-bit machines (size of stack? if the stack
- >is 32 bits, how to work with 16-bit arrays?). If somebody wants to
- >devise a graphic standard (variable names that contain the screen size?)
- >that is also fine. Not all 32-bit machines have graphic interfaces.
- >Not all people want floating point support. The standard should be
- >modular. There should be a 16-bit standard. There should be a 32-bit
- >standard. There should be a floating point standard.
-
- I find it curious that "16 or 32" bits is still such an item of discussion
- here. In most other languages, the question is beginning to be "what to
- do about 64 bit machines"? In this sense, Forth is mired 10 to 15 years
- in the past, which is when most hardware, and other languages, made the
- 16->32 transition (or in some cases, the 24->32 transition).
-
- Currently, applications are growing by about 1 bit of address space every
- 3 years. 32 bits is already too small for large commercial apps, not to
- mention large scientific apps. 16 bits is a joke, except in the extreme
- low-end like video games and the cheaper PCs. (Even there, it leads to
- abominations like segments. Programs deserve uniform address spaces.
- Segments are for worms.) The transition from 32->64 bits can be expected
- to provide adequate address space for most of the 21st century. Then we'll
- probably have to go to 128 (sometime between 2070 and 2120 most likely).
- The starship Enterprise's computers probably use a 256-bit word. :-)
-
- This long period of word-size stability will be an entirely new thing in
- the computer industry. We've never had that before. Maybe we can finally
- get our act together and write software that can be useful for decades
- without vast amounts of maintenance. Or maybe not; we (as a profession)
- have not demonstrated any such ability to date.
-
- There already ARE graphics standards. The nice thing about standards is that
- you have so many to choose from. :-) Postscript is Forth-like, so perhaps
- NeWS or NeXTStep would be easy to interface. But they don't run on every
- architecture, or even most, and probably won't for a long time.
-
- For floating point, could there POSSIBLY be any reason not to support the
- IEEE floating point standard (sometimes known as "The Wrath of Kahan")?
- Not only has it been carefully designed from a numerical analysis viewpoint
- (which 99% of Forth (or any other language) programmers would be incapable
- of doing), an increasing number of machines have it built into the hardware.
- This choice would still leave some issues unresolved, but would go a long
- way toward getting a reliable, portable, robust standard.
-
- Howard A. Landman
-