See what you see. I don’t perceive Rails as a code generator so much as a rich set of objects and methods, with a lot of interesting things going for it (cool as hell progress bars and the not-showy-ajax-but-truly-brilliant-ajax stuff) but I decline to disagree with Steve about that as he is undoubtedly more in the know.

My knee-jerk answer is go with ruby. I love ruby. I coded with it extensively while I was in school and dream about it when I’m slogging through perl’s object implementation, which I use dutifully and with ignorance of the joys that Conway, et al find there. With that disclaimer I can say what my gut feels, and that is that however wonderful someone could find perl they will find ruby a long measure more wonderful. Ruby is perl 6 in a very many numbers of ways and it has been around for a pretty long time now, so the user support is there. Gems works better than CPAN in my personal experience… not as extensive but very quick and painless.

Maybe my suggestion for ruby is a nostalgia speaking — I coded maybe 10k lines in ruby during a 9 month 18 hour a day teeth-cutting, eye opening computer science orgy. So I went there with it: I coded an RSA encryption system, a web application framework, I wrote my own mysql driver, my own multithreaded webserver, etc. In Perl I have written 20k lines that move and rename files in the last six years with some extra credit csv file manipulation… certainly no romance there and a great many embarrassing hacks.

But ruby is all about the romance and the sexiness. It is typographically gorgeous. It allows for all kinds of thinking and styles of coder. It has like 200 kinds of smart going for it. It seems to perform well enough. I really like the implementations of various things. There are really good books about it and a lot of interesting work being done with it.

It sucks a lot less than PHP. But *everything* is going to suck when you know perl so very well, Felix. I don’t even really like perl, but I’ve written 100x more with it than everything else put together, so, like Steve I find myself deciding to start some project in ruby and then not being able to figure out how to do something (like proper process control) that I have done in perl a bamillion times (or a dozen, or once) and then I just back off and do it in perl, cuz, really who has the time? The answer to that just may be that *you* have the time and that you will be able to realize a return on your invenstment in learning and struggling with ruby by being able to work better, faster and happier in the long run. That seems to be the usual case: people move from perl or java to ruby+rails and find religion in a very short time — and that might be the key element right there: it wouldn’t take long to inform yourself enough to make a decision that will satisfy you enough to keep you from fretting about this kind of thing over and over again.

Good luck. (rubyrubyruby!)

As for LISP and Scheme, remember when we did a little Scheme at Capital Thinking, Steve? Hah! That was great, and actually sometime last year I re-picked up Common Lisp. Did a little with it, but I’m not sure that I want to bust it out for this next project of mine.

I also tried learning Haskell, but spent a few days trying to figure out monad’s, talked to Larne who I thought was a big Haskell booster and he dampened my enthusiasm so that kind of tapered off. Ah well.

With PHP, some of my concern is language size and craziness - it’s core system is just so huge. It seems to be the cause of much security concern - I honestly haven’t been paying too much attention, but is that the case? As for purpose built - I really haven’t seen much advantage of PHP’s embedded language over HTML::Mason and have seen much advantage of HTML::Mason over PHP. Granted, I have limited experience with PHP, but if you could give an example? And an online manual’s cool, but I’ve got perldoc -f. :)

I think it is without doubt that PHP has the momentum these days. There’s a lot of interesting work going on there. Java’s momentum is slowing, but honestly I don’t see the death of any language. People keep using them because there really is no one true language. People thought Python would kill Perl and Ruby would kill Python. I mean, people are still doing TCL.

I’m still perplexed, although I admit to a little more PHP leaning again over Ruby, but that changes kind of hourly.

Java bloat: definitely. It just doesn’t scale down. I don’t think it’s dead, though, due to Apache Tomcat/Geronimo. The potential to open up enterprise-class Java to the masses might be analogous to the Apache web server’s effects. I forgot to mention strong typing as (IMO) a major scaling advantage to Java. Anyway, every geek needs food and energy drinks, and Java will make that happen (cash-wise).

Swain: very interesting point about multi-core. I learned SML/NJ and Common Lisp in college, and SML was pretty tricky. Lisp, though, is absolutely amazing. Talk about parsimony. Paul Graham really illuminates this in his “The Roots of Lisp” at http://www.paulgraham.com/rootsoflisp.html

Multi-core is the future of computing, and it’s clear that “our” current approach to software doesn’t really scratch the surface of parallelism. When I write Lisp, though, I depend on indenting to keep my parenthesis in order (I wasn’t taught this - maybe it’s bad style), and then I wonder if Python’s indenting isn’t so bad.

Also, thanks for PHPWiki! I use it, and I’ve had an easy time extending it for my needs. This is an example of a great match of code size versus project scope. However, through no fault of the PHPwiki project, it’s at the mercy of the PHP developers to make a safe container for it.

If you want to read something discouraging about PHP, check out the securityfocus interview with Stefan Esser (linked off of rtfa.net). Well, maybe it’s encouraging, since the article is all about making things better in the future.

Seems like everyone is pretty enthused about Ruby… The O’Reilly Nutshell book (mine is 2002, based on Ruby 1.6) is under 200 pages. Beautiful!

I first wound up coding in PHP way back during the days of PHP 2.x, when I worked at cleveland.com (circa 1996). My experience then was quite positive; it was miles better than SSI, which was pure torture to work with. (Was and still is).

But after I moved on to the NYT, where I could do all the Perl I wanted, I still longed to return to PHP; in 1999 I wrote/released the first version of PhpWiki, written in PHP 3.x. And that project was fun to do.

Over the past few years I’ve looked at Template Toolkit, HTML::Mason, and Embed Perl and there was something that didn’t quite *feel* right about them. I couldn’t put into words what it was until fairly recently. Here it is:

mod_perl is about *extending the Apache server*. But PHP is a domain specific language: dynamic web sites. PHP comes from a procedural language background, but with PHP 5.x they’ve nailed a pretty decent object model; so there’s a rich library of functions (some with pretty funky names; you implode() and explode() strings instead of split()ing and join()ing them, for example); and I think PHP’s true killer app is its online manual. Sun’s Javadocs for Java come pretty close, but PHP’s annotated manuals are the bomb. It’s right up there with CPAN in terms of “something every language should have.”

Yahoo and LiveJournal run on PHP; and it was PHP+MySQL that was the basis for the “shared nothing” architecture. (And Apache; let us not forget Apache). So unequivocally, yes, learn PHP. It’s worth it. If only for throwing together the occasional quickie dynamic page: you can write a Perl CGI, true, but PHP was born as a dynamic web page language and you don’t need to install any modules to urldecode a string. It comes with the language.

Ruby: if you go by O’Reilly’s publishing schedule, then Perl is Anna Nicole Smith and Ruby is Scarlett Johansen. Ruby arrived at the right time and the right place: Perl 6 was announced in, what, 2000? Seven years ago? I fear Perl might be taking its place alongside sed and awk as traditional sysadmin tools, and that will be all she wrote. (Which isn’t so bad. Hardly a day goes by that I don’t reach for Perl to do something one-offish:

perl -pi -e ’s/someOldFunction/someNewFunction/g’ `find . -name \*.php`

for example). The problem I’ve had with Ruby for the last five years (I’ve tried, I think, three times to pick it up) is when I don’t know the immediate language construct to do something (open F, “foo”; @foo = ;) it frustrates me and I think “I can do this in Perl in ten characters.” So Ruby doesn’t get a fair shake from me.

Rails, I think, is fine if you know what’s going on under the hood. In fact you might not be able to do anything serious in Rails unless you do understand all the principles and the code generated; and it is a code generator, and all the wins and travails that entails. Rails hasn’t quite proven itself in scalability yet. You have to use some wonky unknown HTTP server to get any performance out of it.

PHP also came into its own around version 4 with the command line. If you build a site in PHP you can drive it from the command line (cron jobs and such) with the traditional #!/usr/local/bin/php type of thing. So Ruby doesn’t hold that over PHP either.

Java is here to stay, sure; but I really, seriously think Java is past its prime. Many new projects will be started in Java in the next ten years; same with C++. But the heyday is over. From here on out Java is going to be primarily a legacy language. There’s really nowhere else for it to go: Sun went after the enterprise market hard, bloating the language with things that the enterprise needs; and enterprise systems really amount to 5% of the market (and 95% of the money, as it happens). So your common geek won’t build much in Java anymore; I think geeks tend to write things like conversion tools and such in Java moreso than web sites. It’s easier and more portable than writing it in C.

So I’m still waiting on Ruby. I know I can pick it up in no time flat, and Rails looks like fun, and I dig the whole Pragmatic Programmer thing. But in the long run I think every programmer should start looking at Scheme or Common Lisp (I’m teaching myself Scheme currently, Prolog after that) because Intel and IBM are coming out with these 80 core chips and it’s an old truth that functional languages are “embarassingly parallel.” Which means they should translate to multicore chips quite naturally. And I want to grok continuations. :o)

Oh, and SourceForge approved my new project today: LHHReplay, a Perl script (to be a Perl module soonish) that “replays” the output of LiveHTTPHeaders. Totally kickass ad hoc web testing for minimal effort will soon be yours!

~swain

PHP clearly scales - I suspect it suffers from similar problems as perl does - too many inexperienced coders make a lot of not nice code. That’s not a problem with the language, as far as I’m concerned, to some degree it is a strength of the language to allow people to do things they wouldn’t otherwise have been able to do - just don’t ask me to go over that code.

I won’t say I liked Java, but I definitely felt smarter developing Java - I think it has a much more academic feel to it. I’m not realy sure why - to me it’s a workhorse language, it’s pretty good to deal with, does everything, but it just isn’t fun to work in. Plus, for most website development projects I think it’s overkill - too much overhead in developing and resource consumption.

Ruby, too me, felt like Perl doing OO correctly. Which was very comforting to me. I haven’t done much more than some initial exploration programming, so I’m not sure what you mean when you say it’s a moving target. Is the language changing that much? Or what’s changing? Other than that new to me concern, my main concern is that there was a bunch of hype for Ruby when it first hit the scene but tapered off. There seems to be hype again with Rails, but I’m wondering now that other languages are getting on the AJAX tip if the hype will wear off and it will be a super niche language again.

As for python. I just don’t like that language. Can’t really say why, I did a few small projects in it and it just didn’t feel right. It wasn’t even the indenting.. shrug.

It would appear that PHP is able to scale well enough (wikipedia amazes me), but if you know the project is going to be big, then there are better languages for it (e.g. Java). If you like editing code inside dreamweaver, then JSP or ASP will still fill that role.

If you code in Eclipse (highly recommended) and you can pick your language, then as a language, there probably isn’t a single advantage to PHP. If you are bound by your hosting company to PHP/Perl, your options may be limited, but through CPAN, I think Perl comes out on top every time.

Ruby on Rails is very exiting, but I have had difficulty developing for it because it is a moving target, and my web host (a2hosting.com) sometimes upgrades beneath my feet. Ruby is just beautiful, but Rails is a little slippery.

Java is here to stay. It’s hard to write small programs in it, but it is very fast, secure (garbage collection, array boundaries, etc), and scales well in several dimensions (project management, documentation, and speed). Linguistically, it’s weird that it has primitive datatypes like “int”, when there’s a perfectly fine Integer class, but I’m not a Java hacker and it might be really handy.

Finally, there’s Google. They appear to be solidly behind Java and Python. The Google culture might be strange, but they are certainly economical, and mostly un-stupid. Google is as Google does. Or something.

Thanks for the comment!