home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.lang.misc
- Path: sparky!uunet!haven.umd.edu!purdue!mentor.cc.purdue.edu!hrubin
- From: hrubin@mentor.cc.purdue.edu (Herman Rubin)
- Subject: Re: Hardware Support for Numeric Algorithms
- Message-ID: <BxDH7G.8Cn@mentor.cc.purdue.edu>
- Organization: Purdue University Statistics Department
- References: <1de9ckINNfj7@life.ai.mit.edu> <1992Nov6.230030.16637@leland.Stanford.EDU> <TMB.92Nov7233431@orac.idiap.ch>
- Date: Sun, 8 Nov 1992 00:54:51 GMT
- Lines: 38
-
- In article <TMB.92Nov7233431@orac.idiap.ch> tmb@idiap.ch writes:
-
- ........................
-
- >Since scientists and mathematicians generally only have little time to
- >spend on programming they (including myself) are probably better off
- >choosing a programming language that supports safety and reuse (i.e.,
- >a language with GC, runtime checks, polymorphism) and that is simple
- >and regular enough so that it can be mastered fully in a short amount
- >of time.
-
- Since Algol, which ignored much that mathematicians were already doing,
- there has been no language designed with the constructs used in mumerical
- computations foremost, or even particularly important. Fortran was
- designed for small programming jobs, not library functions, but it did
- consider what is needed for numerical computation.
-
- Those who rewrote my code without gotos had to work to do it. I am not
- opposed to using "structured" ideas when they help, but they often do
- not, and figuring out what is happening with a batch of unlabeled blocks
- and break statements is much harder than following a natural goto
- control path. It is the way that mathematicians think, and redefining
- continue to be different from Fortran is not going to help. Requiring
- a function call when a goto is what is really wanted, and hoping that
- the compiler can decide if this is what is wanted, or a call-return
- situation with full context switching, is not going to help.
-
- Also, forcing an unnatural syntax is not going to help. I can, with
- some difficulty, read and write Polish or reverse Polish notation.
- With few exceptions, assemblers are essentially Polish. Lisp is
- essentially Polish except for the "standard" arithmetic operations.
- Requiring the writing of pow(x,y) (or is it pow(y,x)?) instead of
- x^y is forcing Polish without even using its power.
- --
- Herman Rubin, Dept. of Statistics, Purdue Univ., West Lafayette IN47907-1399
- Phone: (317)494-6054
- hrubin@snap.stat.purdue.edu (Internet, bitnet)
- {purdue,pur-ee}!snap.stat!hrubin(UUCP)
-