I don't want to live and develop in a world where everyone "should" use "new generation frameworks".

Please, learn how to use object-orientation, learn how to test code, how to design code. Symfony, Zend and any other frameworks almost never the "best choice" (in long term).

@blerou I'd be interested to hear how a framework is almost never the best choice, in the long term. Also, I'm just going to assume that, when you say "framework", you aren't talking about the foundation of every web application (you always have a framework, even if it's custom). But, instead, you mean a third-party framework.

I'm here at Symfony Live in San Francisco right now, and yesterday we had a similar discussion during speakers dinner. More open source software should base their work on an existing open source framework. It will be much easier for developers to work with the existing frameworks, integration between different software is much easier, so it's a win/win situation. Everybody benefits. End-users won't notice a thing, since the software will still work as it always does.

However, it is a very big step for a project like Drupal to make the step. You may alienate a lot of your currently developer community, because they have a hard time to get up to speed with the way the framework works. In projects like Drupal and Joomla!, you may get away with "hacking" things together (not talking about core here, but about custom extension code), but when things are based on one of the frameworks, you won't get away with simply "hacking" a bit. Sure, you can write crap code that works, but it still has to adhere to certain standards.

I would love to see more projects like Drupal moving towards a new framework-based system, be it Symfony2 or Zend Framework 2. I'm just wondering if that will ever happen. phpBB had the balls to take this decision, let's wait and see if the rest also will.

The resource requirements of Symfony are of a different other then Drupal. And Drupal works at a much higher level.

Let alone that things like mvc are nonsensical when applied to the web so Drupal certainly should not move into that direction.

That's a pretty bold statement. MVC is a solid principle and I don't see why it would not apply to the web. Do you have references (or at least some more substantiation) for your claim that resource requirements for Symfony and Drupal are of a different order (I assume that's what you mean)?

I really like Drupal, but I sometimes get the feeling that many in the community suffer from the "not invented here" syndrome. This is Open Source, people. Third party? As soon as Drupal would adopt Symfony, Symfony would become first party. The community may be a different one and when decisions are made wrt the framework, a bigger picture will need to be kept in sight, but I think that's only a good thing.

Of course, skoop makes some very valid points about the actual transition, alienating existing developers, etc. But from a pure technical standpoint... Intriguing.

I think Drupal itself is slowly becoming a framework on which you can create multiple different applications. If you look at distributions like Drupal Commons, Managing News and Atrium, etc.

PhpBB is more a single purpose (messaging board) tool. Which benefits more from such functionality as people will use PhpBB for one part of the system, the other part of the system might consist of a separate image gallery also based on symphony etc. Then you can benefit from a similar backbone of the system.

Drupal can be all of those things (messaging board, image gallery, etc). So why base Drupal on a reusable framework while it'll be the only tool that you will need.

Besides putting a framework (drupal) on top of another framework (symfony) on top of a platform (php + apache) might not be the best solution performance wise. Isn't this how the Java platform is getting polluted with 20 layers of frameworks on top of each other to make everything so abstract that you don't even remember anymore what you were improving in the first place.

I wonder why there haven't been any full-featured CMS packages based on MVC frameworks yet. By full-featured I mean similar to Drupal, Joomla or Typo3, very modular, lots of features.

I'm more familiar with Rails than Symfony, but all Rails-based CMS systems are a blog + hierarchical page manager + asset manager.

Part of it may be that Rails "plugins" are lower-level than say Drupal modules. There's no single authentication system (there's a new one every few months), no single access management system, no single localization system etc., so interoperability is harder.

That said, it would be great to have a full CMS system on top of MVC, maybe it's just hard to get momentum.

I do not think that the reason for the lack of truly MVC based full features CMS is that MVC doesn't fit them. I think its just because the heritage of all of these successful CMS solutions are quite old, predating the advances in the PHP world over the last years. At the same time starting from scratch if you have such a huge feature set is first not trivial, but more importantly a huge risk. As a result they only benefitted by little evolution, rather than revolution.

However given that, and the history of open source, there must at time point come a disruptive force that changes the numbers. Aka something based on a proper framework that is able to cover todays requirements better. Its only a matter of time.

With cmf.symfony-project.org we are secretly hoping to be that disruptive force in the long term. In the short term we are just hoping to be an alternative for people creating customized CMS solutions.

@Johnny van de Laar: the idea is that the Drupal "framework" is completely replaced. I think we'll all agree that Drupal, at its foundation, is a framework. It has some functionality on top that makes it a CMS. The idea is to reimplement that functionality on top of a more "standard" and modern framework like Symfony. It wouldn't be be "Drupal framework on top op Symfony", it would be Symfony with a layer of Drupal sauce (in some places thicker than in others) to make a feature complete product that matches (and hopefully betters) Drupal 7 in functionality.

@CMS on MVC: Joomla is based on an (its own, proprietary) MVC framework :) But that's not what you meant.

I don't know why that is, but I doubt it is the reason you are giving. Rebuilding Drupal on a framework would be exactly that: filling "gaps" because the framework itself doesn't impose a particular standard, providing "standard" implementations for authentication, access management, etc. The challenge would be to do that in a way that you end up with a feature-complete "Drupal", but implemented in a way native to the framework of choice. That is probably where this is going to fall flat on its face, because the framework of choice for most Drupal developers is... Drupal.

This "migration" would involve a lot of paradigm shifts. Everything would change, not just some parts of the system like it has been up till now, from one major Drupal release to the next. I said this morning to a co-worker, what we really have right now is Drupal 1.7 (purely from an architectural viewpoint), not Drupal 7. It might be time for Drupal 2.0....

Note: I'm not allowed to use semicolons, both a valid interpunction character as well as an esstential part of one of my favourite smilies...

Eelke, I think the answer is on the tip of your tongue:

"I think we'll all agree that Drupal, at its foundation, is a framework."

Drupal isn't a product built on a framework. It IS the framework.

Firstly, I'm very glad that this post turns out to be very popular.

I don't understand why Drupal, a CMS, that is a web app (@Berend de Boer), can not based its development on a well tested, full featured, MVC, new generation framework. Just imagine all the benefits. Drupal will not have to rework on its own components, components that every web app needs such as MVC implementation, caching, forms, i18n and l10n, logging, etc. Why to reinvent the wheel for ORM when you have at your disposal a FREE one like Doctrine 2.

It is about being Drupal while reusing the best from the PHP environment. The people from Drupal will implement and market features faster and the people from Symfony will gain a lot of feedback and patches. It is a win to win situation like @skoop said.

The PHP community will start to raise its own level, because the front line projects like Drupal and PhpBB will be based on solid frameworks like Symfony2. This way starting to use the best practices and implementing the standards.

I think it will be harder in the Drupal side but the results worth it. The Symfony community will be more than pleased to help this to happen.

Drupal is a framework and considerably more seasoned than Symfony. It doesn't use web MVC but PAC which is more like Hierarchical MVC. It is also an aspect-oriented framework. As such in some respects it is ahead of the likes of Symfony and Zend architecturally (I say that as an ex-Zend dev). Drupal's weakness and strength is that it has been designed iteratively to scratch implementation itches and so looks messy, inconsistent and incomplete compared against a top-down designed framework. The discussion in the Drupal camp really concerns stepping back, joining the dots and taking a top level view of the design in a pure framework context with Drupal CMS as an application built on it.

Hi Jonathan

We met in Perú in the Drupal Latino.

Moving Drupal to using Symfony as it's framework is not an option because, like people have said before, Drupal is already its own framework. It has tons of modules (more than 5,000) already using this framework. The Drupal framework manages users, content, permissions, paths, aliases, mailing, querying the DB, building forms, ajax, theming and many other things. There is no significant gain from moving to Symfony with all this codebase in place.

The Drupal framework, is also evolving, Drupal 7 made many changes to many core parts of the framework but it's more an evolution, not a full re-write. Drupal 8 and next Drupal releases will continue to evolve the Drupal framework.

Maybe some parts of the Drupal Framework could use more open source projects, for example Drupal uses jQuery. And there was a remote posibility when Drupal changed its dtabase framework to use Doctrine as its framework, but I think that possibility is now gone. Other parts of the Drupal framework will continue to learn from other open source projects and will continue evolving, but using Symfony I think is out of the question.

Jonathan: "I don't understand why Drupal, a CMS, that is a web app (@Berend de Boer), can not based its development on a well tested, full featured, MVC, new generation framework."

Drupal isn't a CMS or a web app, it's a framework for building CMSes and web apps. If you download and install Drupal, then out of the box you pretty much get the most basic implementation of that framework, but that isn't the thing and the whole of the thing.

Here's some of the ways I've used Drupal, both at MTV UK and on private projects:

* www.mtv.co.uk website
* od.mtv.co.uk website (video on-demand site, uses an SMS gateway for subscription payments)
* uk.viva.tv website
* Video ingest and transcoding workflow (with Final Cut Server)
* SMS moderation tools for text-to-screen broadcast services
* EPG services for digital TV
* SOAP/JSON/RSS services for mobile applications and content syndication
* iOS push notification management
* IT support ticket management
* VOIP call logging, reporting and analysis
* URL shortening API

...I could go on.

Drupal is a framework. Calling it a CMS, or even assuming its scope is limited to being a CMS framework, is a huge underestimation of its capabilities.

Just a thought: can you make orange juice from apples?

I've been using Drupal for almost a year now (used at work - no choice). And I can say that it's high time for it be develop using new generation frameworks. I don't really care if it will be Symfony2 or Zend. The current state of Drupal is really a PITA. It has too many features that are not usually used. It's too bloated that it can only serve a few hundred requests per second and uses too much memory. And even worse is that you need to patch the core part for some weird bug or you need to hook into something. All this could have been resolved if it was implemented using OOP which these new frameworks excels at.

Drupal is a CMS, that is what its was built for. The framework is just a result of an evolution of cleaning up stuff and generalizing other stuff. As such the CMS part is design, the framework is only for various smaller pieces. The end result is that you can of course also use it for non CMS stuff.

Now Drupal does have its unique features, especially its take on aspect oriented programming, which in this variant I am not aware any significantly popular framework (or CMS for that matter) features. If you want to stick with this approach, you will not have alternatives, though Flow3 at least also places emphasis on aspect oriented programming though slightly differently (and AFAIK L3 also has some AOP).

Aside from this, I find the Drupal approach a bit confusing. From what I have heard, each new major version has major BC breaks. Of course BC breaks are not binary, but like I said from all I have heard they are pretty big. So I do not understand why they insist on an evolutionary development of the core. If there is something worth a revolution, there doesnt seem to be much argument not doing it in a major release.

To all those that say Drupal is a framework: You are right! However, there is a difference between a framework and a framework. Drupal is a CMF. It is a framework aimed specifically at Content Management. There is no reason why a CMF couldn't be based on another framework like Symfony2. Symfony(2) is really lowlevel. You could easily extend it to do what Drupal does. Drupal would actually benefit a lot from having a third-party framework underneath it. It will save Drupal a lot of work. It's like Jonathan says: being Drupal while reusing the best of the PHP environment.

I'm wondering how many people here have actually used both Drupal and Symfony on sites > 1million pages a day. The idea of jamming symfony under drupal is laughable from the pov architecture and performance. As for 'next gen frameworks', one of the ZF core contributors now uses Drupal as framework of choice.

@meh: I'm talking about Symfony2 and it's not about jamming Symfony2 under drupal, it is about Symfony2 or its components being part of Drupal.

@Joaquin Bravo: is to nice to hear about you. I mean at least Drupal using the components from Symfony2 or Zend 2. Components that are very decoupled that you can extract and use them without problem.

@Ken: I've pass through that too.

@Lukas: I think that revolution would benefit us all because that way Drupal can join the rest of the projects taken a step forward into better practices, standardization and PHP 5.3

@skoop: Maybe the problem would be the people from Drupal does not know exactly all the benefits and functionalities Symfony2 can offer them. The same happens the other way, the people from symfony does not know exactly the same from Drupal (at least everyone except Fabien Potencier ;)). It would be awesome if the core developers from Drupal meets with the core developers from symfony and discuss in which ways they can help each other. This way everyone wins, overall PHP.

While I can see the benefit of writing a Drupal-like CMS using Symfony, I think rewriting Drupal from scratch to use Symfony makes about as little sense as rewriting CakePHP to use Symfony.

I use Drupal 7 as a web framework specifically because of its aspect-oriented module system, and its excellent theming system, which is an excellent "skinning" api built on top of a templating api.

Drupal has its own form-generation system, and its own database layer, and so on, all of which are designed specifically to work with Drupal's module and theme systems. These subsystems are not really hot-swappable with other frameworks. They take a unique approach, and that is Drupal's strength, not its weakness.