From: Jean-Hugues ROBERT Date: 2004-04-23T21:20:57+09:00 Subject: Re: Ruby on .NET (was Re: A Ruby WishList) Hi, On the other hand... I read all of that and it is rather frightening. These people have VM, JIT and almost type inference! Soon enough they will cross the bridge between statically typed languages and scripting languages (coming from the static side, using "generics"). What their acts tell us is: "we love it when it's flexible (i.e. when type is "weak"), but we are never going to acknowledge it however... 'cause we love speed even more (nobody's perfect)." Strangely enough, I don't know of so much work going on in the other direction: coming from "weak" typing, going to fast code. One may conclude that people trading flexibility for speed are unhappy (and try to fix that), whereas people trading speed for flexibility (read: Ruby users) are happy (not trying to improve speed so much). Reality is more subtle. Rite and Parrot are indications that script languages people want more speed. From all that I am confident that future is for scripting languages. Matz: Don't care about Rite, you are already doing the right thing with Ruby: Promoting scripting languages. Some unhappy C# or Java user will soon provide a VM for Ruby. Go to the next step directly: Provide a syntax whereby source code writers can provide "hints" that will help the compiler/VM to speed things up (you may also want to provide easy access to the AST AbstractSyntaxTree, to help the unhappy people). Yours Jean-Hugues PS: Please don't take that mail too seriously, I am not implying that people that like type safety and speed are unhappy people in "real" life ;-) At 15:57 23/04/2004 +0900, you wrote: >//|Don't get me wrong, all I am saying is that the .net is geared to go >//|places and it already is in the enterprise spehere. The merits of >//|platform that provides a true level playing field across >//languages is >//|slowing showing its strenght in the enterprise. >// >//Ow ow ow. The concept of a Microsoft platform providing a >//"true level playing field" for anything just made by brain hurt. > >Hey, backoff. You probably don't know enough about the CLR architecture, >which is why. If you are really interested in technology, go read >http://www.oreilly.com/catalog/sscliess/chapter/ch01.pdf, you might >understand what I was trying to say. > >Yes I figured that Matz's reasons would be those. Which is good. But I >would always argue for the Ruby being on the CLI - I might even do >something about it, had I been a good enough programmer, but that will >have to wait. Today the CLI/CLR let you write a class in C++ and inherit >and extend that in Visual Basic. Being on the CLI simply makes avilable >to Ruby everything the .Net community has done and thats awesome. > >I must say that the CLR is not ready for true dynamic languages, I >cannot imagine how a closure or a continuation would be implemented on >the CLR. The present C# implementation of yield (Whidbey release - VS >2005) is a 'hack', for want for a better word. > >But if Matz (or anyone knowledgible enough) were to put out reasons for >why the CLR is inadequate, the .net community might be able to respond >and do something about it. > >Matz: > > * I want to enjoy writing my own VM. > > * I don't want to restrict the language design by VM feature. > > The latter is more important. When I design new feature for Rite, I > > don't want to wait other projects (say .NET) to add VM function. > >Matz, don't wait - go ahead and do your own thing. I really appreciate >you for that. >Let me repeat myself: >Branching to the CLR to gives Ruby an oppurtunity to branch from a >lesser known scripting language to a player of equal footing to >soimething like C#. Ruby as a lnaguge is good enough for that. Since the >CLR is inadeqaute in ways to support the functionality that Ruby >requires, document them - that gives others an oppurtunity to work to >fix them. > > >There are some very smart people at MS and outside working towards these >things and it Only Helps Ruby to leverage these things. I don't know how >you guys 9especially those who don't know much about the CLR will relate >to this, but atleast notice that there is work happening: >http://research.microsoft.com/projects/ilx/absil.aspx >http://www1.elsevier.com/gej-ng/31/29/23/89/27/31/59.1.006.pdf >http://research.microsoft.com/projects/clrgen/generics.pdf (an old >story, this is part of the CLR now) > > >Regards >Roshan James >http://pensieve.thinkingms.com/ ------------------------------------------------------------------------- Web: http://hdl.handle.net/1030.37/1.1 Phone: +33 (0) 4 92 27 74 17