|
Prev: ANN: SymbolicWeb v0.1 (quite alpha) w/ source code this time
Next: Greenspunning ML (revisited)
From: Spiros Bousbouras on 7 May 2008 12:35 On 7 May, 00:03, Robert Uhl <eadmun...(a)NOSPAMgmail.com> wrote: > Microsoft Windows is to > computing as the Ottoman Empire was to late nineteenth century politics: > it covers a lot of territory, but no-one takes it seriously anymore. He he , this goes to my quotes file.
From: Will Schenk on 7 May 2008 13:55 On May 3, 12:21 pm, "Peter Hildebrandt" <peter.hildebra...(a)gmail.com> wrote: > I believe the barrier to entry is too high here, and I think the major > reason is that the lisp world is so pluralistic: Which implementation do > I use? Which IDE? Which libraries? Where do I get what? > (Unfortunately) people expect there to be one way to do things, i.e. they > expect to go to lisp.org, "click here to download", double-click the > installer, select "example-1" and first launch, click "run", and look at > their first own weblog :-) > > As a newcomer (I remember!) lisp is quite confusing: which implementation > to use? Where to download? Where is a good discussion board? What are > the libraries? I was actually asking myself this question this weekend: I have a code problem which for various reasons I need to break out into something better than ruby. I need local performance specifically and I'd like to make it distributable. (Something along the mapreduce model probably.) I need fairly tight control over memory (at least I need access to mmap() and ideally there's already a good btree implementation that's fast). Event base concurrency would be nice. (Also, it should go without saying that it supports sockets and http.) I know java the best and it's performance quirks, so its a low risk thing for me, but, ug. The ruby code makes extensive use of meta programming, and for what I'm doing I expect to do a lot more. So another reason to not go back to java. I've never used lisp but it occurs to me that this is probably what the hubbub of macros is all about. I've been looking around and everything that's out there is either out of date or lacking consensus. Developing on OSX and deploying on gentoo linux. Everything is so fragmented and incompatible, I mean "pluralistic", that its unclear. If I use Ready Lisp, will I be committing to deploying on SBCL? Do I need to care about that? I want exactly the go to lisp.org and click download this. Ruby has like 5 distros now, MRI, Rubinious, JRuby, IronRuby, some other thing built on a smalltalk VM -- and when I see that many options, I don't see choice. I see: In Someway, Every One Of These Choices Profoundly Suck. I mean, diversity and variation are only positive when you are talking about surface frivolities; core things are shared and remained unchanged. There's not a lot of variation in mitochondrial dna. Lisp seems cool, and I have a good reason now to actually check it out with a real problem. What would you guys suggest to download and install? Pls keep the caveats to the minimum. Thanks. -w http://benchcoach.com
From: Rainer Joswig on 7 May 2008 14:43
In article <53682fc0-b338-4804-be77-84f73a466f00(a)f36g2000hsa.googlegroups.com>, Will Schenk <wschenk(a)gmail.com> wrote: > On May 3, 12:21�pm, "Peter Hildebrandt" <peter.hildebra...(a)gmail.com> > wrote: > > > I believe the barrier to entry is too high here, and I think the major � > > reason is that the lisp world is so pluralistic: �Which implementation do � > > I use? �Which IDE? �Which libraries? �Where do I get what? � > > (Unfortunately) people expect there to be one way to do things, i.e. they � > > expect to go to lisp.org, "click here to download", double-click the � > > installer, select "example-1" and first launch, click "run", and look at � > > their first own weblog :-) > > > > As a newcomer (I remember!) lisp is quite confusing: �which implementation � > > to use? �Where to download? �Where is a good discussion board? �What are � > > the libraries? > > I was actually asking myself this question this weekend: I have a code > problem which for various reasons I need to break out into something > better than ruby. I need local performance specifically and I'd like > to make it distributable. (Something along the mapreduce model > probably.) I need fairly tight control over memory (at least I need > access to mmap() and ideally there's already a good btree > implementation that's fast). Event base concurrency would be nice. > (Also, it should go without saying that it supports sockets and > http.) I know java the best and it's performance quirks, so its a low > risk thing for me, but, ug. The ruby code makes extensive use of meta > programming, and for what I'm doing I expect to do a lot more. So > another reason to not go back to java. I've never used lisp but it > occurs to me that this is probably what the hubbub of macros is all > about. > > I've been looking around and everything that's out there is either out > of date or lacking consensus. Developing on OSX and deploying on > gentoo linux. Everything is so fragmented and incompatible, I mean > "pluralistic", that its unclear. If I use Ready Lisp, will I be > committing to deploying on SBCL? Do I need to care about that? It is good to ask here. You will get many different answers. ;-) Then you still have to choose, but with more knowledge. ;-) > > I want exactly the go to lisp.org and click download this. Ruby has > like 5 distros now, MRI, Rubinious, JRuby, IronRuby, some other thing > built on a smalltalk VM -- and when I see that many options, I don't > see choice. I see: In Someway, Every One Of These Choices Profoundly > Suck. I mean, diversity and variation are only positive when you are > talking about surface frivolities; core things are shared and remained > unchanged. There's not a lot of variation in mitochondrial dna. > > Lisp seems cool, and I have a good reason now to actually check it out > with a real problem. What would you guys suggest to download and > install? Pls keep the caveats to the minimum. Well, many who don't want to struggle (and have the money) use a commercial version (LispWorks, Allegro CL, ...). I would recommend both of them. IMHO both are excellent platforms. For those who don't have the money, like open source, or have some other reasons there are a several choices with some tradeoffs. SBCL: generally fast, largish, threads on some ports (Mac OS X) are 'experimental', widely used. I would just check if it runs on your platform and try it. CMUCL: generally fast, largish (I use it on Mac OS X) Clozure CL: mostly fast, slower FP code, fast compiler, smaller code ECLS: recently gained more traction, good integration with C (embeddable) CLISP: a bit slower than some of the above, small code, quite portable and very useful for scripting tasks The IDE for those is usually SLIME/Emacs. The commercial CL systems have better (IMHO) IDEs - you pay for that. Plus there are some other options. Scieneer CL for example might be useful when some performances requirements would make multi-cpu machines necessary. Corman CL runs under Windows and is useful, when you want to write either Windows software or you want a 'cheap' IDE under Windows. Some old-timers use MCL on Macs. Some math packages run under GCL. Dan Weinreb had a good recent overview: http://common-lisp.net/~dlw/LispSurvey.html > > Thanks. > > -w > http://benchcoach.com -- http://lispm.dyndns.org/ |