PHP5 Vs. CakePHP Vs. RubyOnRails? 469
OldJavaHack writes "If you could start a website (with MySQL for persistence) from scratch and you had a choice of PHP5, CakePHP, or RubyOnRails — which would you choose and why? Things to consider in your decision: 1. Maturity of solution; 2. Features; 3. Size of community of skilled users (to build a team); 4. Complexity/ease of use (for neophytes to master); 5. Greatest strength of your choice, and the greatest weaknesses of the other two. Here is a comparison of capabilities."
Isn't this just asking for flamewars? (Score:2, Insightful)
Brrrr... (Score:2, Insightful)
JSP and ASP.Net (Yes, I know, this is Slashdot) are, IMHO, much more much more powerful and pleasant to work with. If I had to pick amongst the proposed solutions, I'd pick RoR if only for the fact that Ruby is a nice language. I don't know about Rails' features but I could at least trust the language it's built on.
I would start a News website built upon... (Score:1, Insightful)
Of course, that's just me.
the answer: it depends (Score:5, Insightful)
Any halfway skilled programmer will be able to do useful work with any of those frameworks fairly early on, but all of them are also very rich environments, so there's always more to learn.
I've written web apps in an ungodly tangle of PHP4 and PHP 5 and Perl and using Ruby on Rails. Currently Ruby on Rails is in favor, but is far from perfect.
Probably most of my frustration with Rails and PHP 5 has to do with Active Record. My big gripes are: (1) Schemas, entity-relationship diagrams, and queries tell me how an application works -- with Active Record this information is strewn across a whole bunch of files (especially in Rails); (2) Database-independence is a nice idea, but in reality, how often over the lifetime of your website will you migrate to a different database? Usually your database is chosen for you. Usually a switching databases involves coordinating with a lot of people who you'd usually rather not have to deal with -- those issues will take far more time and energy than differences between MySQL and Oracle; (3) a pretty common design pattern for web pages is to have a form that let's you fill in a few parameters (date, maybe geographical information) into a huge multi-table select statement -- you can do that in Active Record, but basically all you gain is a marginally fancier wrapper than you would have with DBI.
Re:Brrrr... (Score:5, Insightful)
It'll change the way you think about development for the web.
Or, if you're really set on Java, try Rails for Java Developers [amazon.com] and you'll see how much more concise the exact same code is in Rails.
Re:Rails (Score:2, Insightful)
Re:Sure (Score:4, Insightful)
Seeing as "Our team is familiar with..." plays no part in this decision whatsoever, I'd say that we are dealing with kids writing Their First Site. Or, looking at the peculiar phrasing and noting the date, kids who have just received their first homework assignment for a project that requires them to submit a plan first.
Re:Errr, this is a new story (Score:5, Insightful)
Lately people (aka: script kiddies) seem to be losing the distinction between what is a language, and what is a framework. I cannot remember the last time I downloaded a PHP script and it required PEAR. I absolutely despise PEAR, and all other frameworks that really don't seem to have a place.
Over the past 5 years or so (I develop websites for a living) I've developed a framework-style setup that I use for all new projects. Most sites don't share the same code as I develop project-specific. But the structure is the same, and in most cases I could grab a pile of files from one site and plop them in the next and it would work.
Use the tool as it is meant to be used. PHP is a language. A framework is a framework. Please don't compare them on the same level.
Wrong Criteria, Wrong Problem, Wrong Solution (Score:5, Insightful)
You've provided no information on the actual website that you intend to develop. That's the important part -- the features and functionality to the customers and end users.
Instead of considering the features of the language and framework first, how about the features of the application? How many users? Who will be supporting it? What kind of server resources are available? Do you need internationalization? What's the roadmap for the site over the next 3 to 5 years? Maybee then you can map the features of the website to the features of the framework or language, such as the maturity of the libraries directly related to your webapp.
But picking the implementation language independent of the functionality of the website is a classic sign of solving the wrong problem. I don't care what you program it in, if you're asking these questions first, you are programming it in the wrong language.
Tapestry (Score:4, Insightful)
Rails just does not have a stable server. Webrick + fastCGI, or Mongrel, they both crash regularly for us. Also I've had to maintain several Rails apps written by others, and it sucks. All those neat tricks that makes it "productive" for the first programmer makes it difficult to understand and maintain for everyone else.
In other words... (Score:5, Insightful)
I've never used Ruby or RoR... my experience with PHP is limited as well...
In other words, you were trolling. :-)
Having done websites in PHP, Rails, Python and Java, I can say that they all suck one way or another. Ruby and Rails are both very different from PHP and my personal unconfirmed suspicion is that a lot of the Rails problems people have are from programmers who jump over into Rails without first learning what they're getting themselves into. Deploying Rails can be very difficult and you can face a lot of issues that you would never face for PHP.
Personally, I prefer Python or Ruby over PHP any day.
Have your cake/php/rails and eat it to (Score:4, Insightful)
What we have here is another usual question that all really depends on your project type. That being said, I'll try to break from the typical, slashdot format and attempt to address your question:
P.S. A similar question of Rails vs PHP vs Java question was somewhat subjectively discussed late last year http://www.cmswire.com/cms/industry-news/php-vs-ja va-vs-ruby-000887.php [cmswire.com]
Re:Doesn't it depend on what you intend to do? (Score:3, Insightful)
And, yeah, comparing Rails to PHP5 is kinda dumb because Rails is a framework and PHP5 is a language. The problem with using "PHP5" in any non-trivial application is that you invariably end up building your own framework(s) which can be a huge time sink. Now, comparing Rails and CakePHP would make much more sense. I haven't used CakePHP, but it looks like it has all the major framework features that Rails has. Though being based on PHP is a huge handicap, IMO. PHP is just kind of a dumb language... at least compared to Ruby. Ruby is just an awesome little language. If only it were faster.
Re:the answer: it depends (Score:3, Insightful)
If you told me I could pick either the database to use or the scripting language to code in, I'd pick Postgresql and let you pick the language. Most of the things people try to do in scripting languages can be handled in the database much more elegantly and scalably. Of course, most people don't realize this because they've only used MySQL and don't realize how much it's missing.
If you told me I could pick both, I'd go with Perl, unless I were doing something very simple that's been done a thousand times before.
PHP is decent enough for what it is. Historically there have been security problems with it, and the design is crappy. But it's quick and easy.
I've never used Rails, but it sounds like it does a lot for you. In my experience, that can be a blessing if you're doing something textbook. If not, have fun fighting with the assumptions other people made.
Re:Brrrr... (Score:5, Insightful)
public class DoStuff {
protected double someNumber;
public setSomeNumber( double number ) {
try {
someNumber = number;
} catch (Exception e) {
}
}
}
... with some PHP for the same code, which would look like this:
public class DoStuff {
private someNumber;
public setSomeNumber($number) {
try {
$this->someNumber = $number;
} catch (Exception $e) {
}
}
}
I don't see how that's wacky syntax in the slightest. Just people people use PHP like it's Perl+Mason doesn't mean you can't use PHP for serious, scaleable, enterprise software. I know from experience that people are just as likely to write nasty Perl, Ruby or ASP as they are nasty PHP.
Personally I think Java makes it more difficult to be wacky (even though of course it can't force people to write code that's ultimately good) and that has definite benefits in an enterprise environment, but that lack of flexibility (which scripting languages like Perl and PHP have) is also why I don't tend to want to use Java.
Re:Errr, this is a new story (Score:3, Insightful)
Except that in this case, the comparison is valid. PHP without a framework is, if you will, the null framework. Its advantages are that there is no learning curve, deployment is trivial, and there's no runtime framework overhead to heat up your CPUs. On the other hand, it doesn't help you at all in writing the app, and once your site gets beyond a certain size you'll probably end up writing a framework anyway - as it sounds like you've done, after a fashion.
Re:Sure (Score:4, Insightful)
The beauty of Ruby is, even if you don't like it the way they do it, you can always monkey patch it. Open up the object and override the method(s) you don't like.
Try doing that to the PHP core libs. Better know C, and love it a lot.
It makes little sense to say Rails doesn't scale.. (Score:5, Insightful)
Granted, I wouldn't write Digg in it, but *I'll never write Digg in anything*. Neither will 99% of the world's programmers, and for the 1% that are making social networking sitse with desired user numbers the size of nation states, they have the LAMP stack and God bless them for it.
As for me, I've got one quite profitable desktop application written in Java (folks laughed at me for that -- what can I say, it got the job done) and am having a bloody ball working on a small business vertical app which, at $15 / account / month and low predicted need for users to interact with the app, would replace my day job income at about three dynamic page hits per hour. I have this funny feeling that Rails will scale that far.
If you've used Rails, then CakePHP /hurts/ (Score:4, Insightful)
This is not to knock CakePHP. In its own right, CakePHP is an excellent framework and a lot of quality work has gone into making it what it is. It's a powerful framework.
The move to this kind of framework can be quite a mind job, whether you're moving to Rails or CakePHP. It requires breaking down very solid foundations of ideas that you've built up over the years on how to build a web application. If PHP is your thing, then weathering that mind job will be all the more easier if you're doing it in a language already familiar to you. But if you're willing to try something new, then it's worth making the jump to Ruby on Rails.
Re:Brrrr... (Score:2, Insightful)
Java is great if you work on a single application with its own hosting environment, if you need multiple websites or webapps it quickly becomes a configuration nightmare. I think that is why a lot of experienced java developers seem to really like Ruby, because right now configuration is the single biggest hassle in the development of java websites.
Re:Brrrr... (Score:3, Insightful)
Re:Python and Django (Score:3, Insightful)
Yes, damn that Python. I mean, Ruby is so much cleaner. In Ruby, everything's an object! Wait, no, that's true in Python too. But in Python you have to prepend self to talk about an instance variable, that's ugly! Oh, wait, Ruby uses @ and has self for some cases, too. Hm. Well, Ruby's got all that metaprogramming goodness, surely you can't do that in Python! Er, no, forgot about __metaclass__ (which, by the way, doesn't have an equivalent in Ruby -- you have to monkeypatch the class object after it's created), never mind on that one.
Etc., and so forth. Only a "fanboy" would proclaim one language or the other as being inherently superior; in the long run, the biggest difference between the two is idioms (the languages themselves are remarkably similar once you discount superficial syntax differences; see, for example, Python "fanboy" Alex Martelli explaining how Ruby and Python are far more similar than different [google.com]). And in doing so you've tipped your hand.
Re:Brrrr... (Score:3, Insightful)
Re:Sure (Score:5, Insightful)
You could, you know, link to those "way too many blogs" and thus let the rest of us decide for ourselves if this is incriminating evidence against Ruby.
"I read it on a blog" does not in any way imply truth.
"I read it on many blogs" doesn't really make it much better.
And until then, you shall remain a troll. After you post the links, you'll have your status upgraded to "person with an opinion, willing to discuss".
Not hard (Score:1, Insightful)
One of the most visited sites with dynamically generated content, Wikipedia, is written in PHP. So, I'd say that shows it's scalable and heavy-duty.
Re:Python and Django (Score:1, Insightful)
ROFLMAO
PHP beats RoR on deployment (Score:4, Insightful)
A key issue, in contrasting PHP with RoR, is deployment.
Deploying PHP is easy in most environments, perhaps as much because of its age as because of its inherent character. I work in an academic environment, in which all professors and students have the ability to make PHP sites. Each of my personal computers also lets me make PHP sites with no difficulty. Deployment amounts to no more than a file copy, perhaps with a change of file permissions. (I won't mention the database work, because of course it is the same for all schemes, PHP, RoR, etc.)
But, unless you're using a host that has been set up to server RoR, deployment may involve changing Apache configuration files, compiling new Apache modules, etc. Such changes require root access (not available to folks sharing machines), and have the potential to break the other sites on the machine.
I think there is a reason why the RoR tutorials, books, and promoters so seldom mention deployment: it is difficult for many people in non-commercial environments that are not set up for RoR.
Oh, and one more thing. All of this fiddling with apache is boring to those who have set out to create websites. Learning Ruby to do RoR is quite fun, actually, and it has the advantage that it lets you use Ruby for other tasks as well. But learning apache doesn't help you with anything but apache; it's a bit of a single-lane road.
RoR has a sort of elegance about it, and you gain a great deal of functionality from the system (e.g. for logins, etc.), and so it is a terrific tool for rapid development, particularly of an evolving idea for a site. It sounds crazy, but the optimal path may be to write the site in RoR and then rewrite it in PHP, so that deployment will be easy and so that the site will scale well[*].
* -- I've not mentioned scaling and speed because these issues are covered in other posts here. Basically, RoR is not impressive on either.
Re:Sure (Score:2, Insightful)
PHP users seem to spend inordinate amounts of time worrying about the latest exploit in the PHP engine, or how easy it is to introduce exploits at the application level thanks to the awful library functions, while RoR users end up mulling over the lack of scalability in their apps once the initial prototype is complete. Meanwhile, "boring" Java apps continue to power large scale systems with increasing reliability and maintainability. Perhaps it's because neither PHP or RoR has the depth of experience or literature on best practices that J2EE now has. That's not to say that things have always been so rosy in the Java world - EJB conventions buggered up a lot of projects until a decent set of patterns were identified and people had bought into the ideas behind lightweight frameworks such as Spring.
Language comparisons (Score:4, Insightful)
PHP is just ugly, (see grandparent) Perl and Ruby are quite fun and share the same philosophy (see TIMTOWTDI [c2.com]) where Python is just the opposite (although fun too).
Python and Ruby also share their deep roots in clean object orientation, where Perl's OO syntax is - though bolted on - very flexible and even more TIMTOWTDI than Ruby's.
Just to set things straight, I really like Perl, Ruby and Python, though I can confirm that developing web applications with Django is bliss.
Re:Nriyh (Score:3, Insightful)
It's difficult to understand, since PHP combines the quirkyness of Perl with the syntactic bloat of Java and does have some other shortcommings. However, it handles all web stuff very gracefully. Tons of functions in PHP are built to handle the everyday shortcommings the WWW Inet-service brings along. And they are all come in the box without any classpath or you-need-this-seperate-servlet-running-to-handle-
PHP is the king of all SSI solutions because it's built for exactly that sort of task and nothing else. That's why the most successfull, non-trivial web applications are built with it. It's the best tool for the job, plain and simple.
Static typing (Score:1, Insightful)
Twitter Follow-Up (Score:4, Insightful)
And there's a good follow-up [romeda.org] by one of his coworkers:
We've been extremely happy with Rails, and make use of the multitude of helpers that it offers us - like any application on any stack, though, providing fast response times to a (rapidly) growing number of users is a challenge. The solutions are often tightly coupled to the application and its characteristics, and while scaling the most trafficked Rails site in the world, we've run into situations where existing solutions weren't enough.
Rails is best at database baby-sitting, which is not what Twitter is about and it's understandable they would have issues. Ruby is slow and we need a good virtual machine. Nevertheless, Twitter does run on Ruby which shows that it can be made to scale. Not that Twitter is a good measure of anything other than, well, Twitter. And I'm sure someone could have done it with PHP, Python, Erlang or C.
Which is always why blanket statements about languages and platforms is always a bad idea. Just look at the comments on this article. It's just a chance for everyone to trumpet their favorite web framework or language. Sure we have our favorite tools, but most of them suck at one thing or another.
I learned PHP once (Score:4, Insightful)
I started learning it. By about a chapter into the PHP book, I was thinking "holy crap, this language is uglier than perl". It has everything you would expect from a language thrown together by people who were either ignorant of software engineering or aware of it, but aggressively hostile to it. Everything global by default? WTF?
I have never seen a language with so many carefully crafted security holes that the developer needs to learn to avoid. Default behavior for inclusion is to allow URLs, so you can, you know, run code from any site in the world. There's a feature everyone always wanted, which is never going to be subverted!
I made it through about two and a half PHP books. In that time I learned that the MySQL and PostgreSQL interfaces were substantively different, and of course, used differently-named functions with slightly different calling conventions. Why? Because there's no abstraction or generalization going on; just whatever features sound cool getting thrown in with some name that wasn't previously in use. I learned that this is just BASIC all over again.
I spent several days thinking hard about bleach, and went back to programming languages that were designed with some kind of consideration given to the development of larger projects.
Ruby's undoubtedly "slow". That's what everyone said about perl and awk, too. Come to think of it, I've had people tell me that C was too slow. But Ruby has the amazing, shining, virtue that it is not a stupidly-designed or ugly language. I spent a while working with Ruby, and some helpful people pointed out that, in fact, the language does have a gotcha to watch out for. One. Not so many that you have to buy whole books full of things that you'd obviously try that don't work, open your site up to XSS, or behave erratically. No, just the one.
Can PHP work? Sure. But the tacked-on afterthoughts provided to allow you to, in theory, if you remember to and want to put in the work, use basic software engineering principles, are not enough. The language provides a huge array of runtime functionality, with a function for everything. It doesn't provide the basic tools you want for engineering large projects, meaning that the workload of maintaining big stuff in PHP is exponential, not just quadratic.
It can be made to work, but it really is that badly considered, and I wish people would stop doing things in it. Life is easy enough for the botnet people already, we don't need a language in which you have to be warned not to set the flag that lets remote sites set every global variable in your program.
Re:Sure (Score:3, Insightful)
Re:Sure (Score:3, Insightful)
I had the same experience. Right now I'm trying out Catalyst [catalystframework.org] + Mason [masonhq.com] running under mod_perl.
Catalyst takes most of the good ideas from RoR and combines them with Perl's TIMTOWTDI philosophy, rather than Rails' "our way or the highway" attitude.
Mason is a PHP-like templating language for Perl.
Re:Sure (Score:1, Insightful)