From ... From: Erik Naggum Subject: Re: Best Lisp for Windows NT? Date: 1999/05/05 Message-ID: <3134911590693621@naggum.no>#1/1 X-Deja-AN: 474378392 References: <372c9359.702572255@news.earthlink.net> <372de455.723304977@news.earthlink.net> <3134715279176341@naggum.no> <372e9030.832835754@news.earthlink.net> <3134816211797975@naggum.no> <372FB204.325B1E9B@mindless.com> <3134865083642662@naggum.no> <372fff1b.926766920@news.earthlink.net> <3134899638508182@naggum.no> <373155f4.948999168@news.earthlink.net> mail-copies-to: never Organization: Naggum Software; +47 8800 8879; http://www.naggum.no Newsgroups: comp.lang.lisp * please-reply-in-the-forum@thank.you | Ok, once they get to know you, to be sure you're a genuine prospect, they | let you do an extensive evaluation. Good. But, before they get to know | me, I want some kind of evidence that their product is even worth | considering at all. I want to see a bunch of sample programs and the | resulting executables. I have a hunch the executables will be too big. | I'm not seriously interested in ACL5 yet, and won't be until that hunch | is proven wrong. They obviously only want to give evaluation copies to | serious prospects, and I'm not one yet. as I have already said, the problem is getting to know Allegro CL (or any Common Lisp environment, actually) to begin with. I recognize your issue. I just don't think you're approaching it in a constructive way. incidentally, I don't think hunches can be proven either right or wrong. I think hunches are good because they allow me not to waste a lot of time thinking about something that my gut feeling rejects, so I expect your hunches to be on my side if I'm selling Common Lisp and you approach me. if you approach me with a bad hunch and want proof your hunch is wrong, I don't think I have the time (or inclination) to counsel you out of it. personally, I don't think executable size matters at all, so I'm obviously not being helpful to your particular quest, but let me show you just how little executable (actually, image) size matters to me: I discovered that by precomputing the decimal representation of five-digit numbers and inlining their printing, my whole application sped up by a factor of 8. that's a 500,000-character string just wasting space. I had already precomputed the break-out of 400 years of calendar data and 86400 seconds of time data, which added another megabyte to my image, but those, too, were very serious space/time tradeoffs. it is helpful if you consider that any proper function returns the same value every time it's called, and so can be precomputed, at potentially huge space costs. I have been willing to dispense with 1.5 megabytes of memory in exchange for very close to zero cost in some bottle-neck operations. | Right now, what I would recommend Franz do to make their product more | popular, would be to put a bunch of sample programs on their website, | with the Lisp source code and with executables, DLL's, components, etc., | built from that source code with their product. People need to see the | results, even more than they need to see the development environment. _which_ people would that be? | If a programmer can show his boss a program, run it and show that it | looks like any other program, and how how big it is, and that it will run | on different computers, and then say to his boss, that program was built | with ACL5; that alone could go a long way towards getting Lisp more | accepted. I honestly don't think so. I do think there's a place for Common Lisp in the low end of the market, I just don't know how to approach that market, and I'm worried that the Lisp vendors will squander their efforts at trying to attact people who think Bill Gates is a hero and his products are safe investments, or, even worse: the people who think he's a criminal and his products are shit, but _still_ buy them and expect others to take the blame when they hit the fan. #:Erik