Re: Mathematica language issues

*To*: mathgroup at smc.vnet.net*Subject*: [mg53151] Re: Mathematica language issues*From*: josephoswaldgg at hotmail.com*Date*: Sat, 25 Dec 2004 04:00:58 -0500 (EST)*References*: <cqbh9b$4k4$1@smc.vnet.net>*Sender*: owner-wri-mathgroup at wolfram.com

One risk of saying that expressions are "unlikely to have any practical use" is that sophisticated programming evironments tend to be used to automate tasks that are relatively unrestricted. Perhaps no human using Mathematica would type in these expressions by hand, but it is certainly conceivable that someone might develop an application package that would end up generating forms like these as intermediate steps while addressing very sensible and practical problems. Think of something like a C language compiler: it is clear that C compilers will, even for simple, straightforward C expressions, generate machine code that is quite unlike what a human machine code programmer would create. If a CPU acted strangely because its designer never anticipated codes in these orders, people would complain that the CPU has bugs. Perhaps compiler writers would learn to work around these bugs, if they were isolated and predictable enough, but these would be bugs nonetheless. People trying to build large systems need building blocks that are *robust*, in the sense of behaving predictably *even in situations that might not be forseeable*. You might get a warm fuzzy feeling from experimentation and discovery, but people using libraries will end up performing mysterious uncontrolled experiments, not of their own choosing, and perhaps silently returning bad results. --Joe

**Follow-Ups**:**Re: Re: Mathematica language issues***From:*Andrzej Kozlowski <akoz@mimuw.edu.pl>