I'm trying to identify some of the pros and cons of having a CMS that is event driven.
Event driven is not uncommon. You see it in many scripting languages like Actionscript, javascript, jquery that involve a client. How about in a CMS where the events and their responses happen on the server. What advantages or disadvantages might this approach have, and what other approaches are there that people may prefer more.
P.S. Please note that I use Actionscript, JQ, and JS as an example only. You realize that when talking about a CMS this way, the events an their responses are all server side stuff.
Edit: I see a lot of people saying that it makes no sense to use event-driven as they don't get what it is. One of the CMS systems that already use this approach is Drupal, so trust me it's an existing way, I'm not pulling ideas out of my A. It just means the "internals" of the CMS (all the server-side stuff) are event-driven. The core does its thing AND defines events. Plugins can respond to those events to add their own logic. I mentione Actionscript as an example because client-side is where this concept is most known, but it can be on the server-side too, just maybe not as relevant for normal applications and so not as known. But it makes sense for something more complex like a CMS where other developers want to add their own plugins or even change the pre-built in logic of the CMS.
Are you sure "event-driven" is the correct term for this?
I think what you are talking about is a plugin hook infrastructure that allows plugins to act when an event takes place (a hook is triggered).
What I know as "event driven" is when desktop applications store events in UI elements, just as Javascript can do for HTML elements. Desktop apps are built entirely that way. That can never be achieved in Web-based PHP because it is entirely request oriented.
Anyway, I see what you mean now. There are CMSs and frameworks that have this to some extent - Wordpress for example and Dokuwiki.
In Wordpress, these events are called actions. You can see a full list in the Wordpress Plugin API.
In DokuWiki, they are indeed called events: DokuWiki's event system
In addition:
I'm trying to identify some of the pros and cons of having a CMS that is event driven.
The pros are obvious: It becomes much, much easier to hook into the system without having to hack the core. It becomes possible to write real plugins.
One big con is that the system tends to become slower on the long term the more hooks there are, and the more plugins are registered to the hooks. I've seen a huge portal's maintenance operation - the deletion of several hundred Drupal nodes - take hours on a ultra-fast production server, mainly because of the hook system.
What exactly do you mean by "event driven"? Does it mean, that clients can watch content and will be notified, when it is updated?
If so, than this is really a great idea, but the implementation is an ambitous undergoing. The clear disadvantage is, that there are tons of things, that can go wrong, while a request based model is much simpler and thus more robust.
Related
Having been dong research on codeigniter,kohana and fuel php, they seem excellent for the large scale project I want to create with various sections of code. The system I want to build is to be a core set of code, which can produce a full ecommerce web application with plenty of modules, or can produce a simply few page site, with a news or gallery or whatever.
I have started thinking that I may be looking at the wrong thing for producing basic sites. From my very brief research, it seems (atleast on codeigniter) I need to use the 404 overwrite to point to my pages controller in order to handle basic content pages (from the base url /).
Should I consider a seperate code set for the basic sites, and use these frameworks only for large web apps, or is this the normal way frameworks would handle serving basic content pages.
I personally have a core framework (MVC/CodeIgniter) which I use for even the simplest static sites - Because someone will always say "Oh and can we have a shop" or a contact us form or ...
The overhead of the frameworks themselves is negligible if they're not doing any work - so why not put them in?
Another bonus is that you now know exactly what environment your modules will be in - you don't need to worry about maintaining one versions for sites with the framework and one for those without (One other benefit of using a framework is that it makes if far easier to maintain code in general - especially in multiple-developer environments.)
Th templating engines provided by most frameworks (or easy extensions to them) are often useful even for static sites (custom controls like a news ticker or custom markup/js validation for certain controls)
In short, unless you have extremely limited server resources (mobile device?) use a framework
Edit:
I'd add a note of caution - PHP is a very flexible, powerful language and when used well can create some truly fantastic sites. When used poorly, however, it is easy to make insecure websites.
From your post, you seem relatively new to PHP/Frameworks. Can I suggest you make sure that the framework you pick enforces good behavior (as much as possible)
CodeIgniter removes unsafe $_POST and $_GET variables, replacing them with safe equivalents. It also provides a parameterised mechanism for querying the database which helps avoid SQL inection attacks. These are important things to consider when comparing frameworks.
I personally found CodeIgniter to be a good balance of maintainability, security, extensibility and functionality.
I would consider using a CMS for the basic sites, something like Drupal, they do a lot out of the box and should satisfy your requirements for a basic site.
As for the eCommerce part, from experience its best use something that is separate from the CMS. Largely because projects or frameworks that are designed for that sort of thing are generally an application.
CMSes do what they do they do best, but that all they really do :)
Also have a look at Magento for your eCommerce needs.
Hope this helps.
I think your question is about what to use for "basic websites".
Should I consider a separate code set for the basic sites, and use
these frameworks only for large web apps, or is this the normal way
frameworks would handle serving basic content pages
If by basic sites you mean websites that have little to no dynamically drawn content, I would suggest a CMS such as Wordpress. You mention that you are using PHP and Wordpress is a very popular option. If you are creating a blog and a very static pages, this will be a great option. It can even be modified to do more 'dynamic' things that can be considered "CMS Territory". Once you get the basics of Wordpress down, and if you still feel that you need to modify it a lot, then you should look in to a framework that will allow you to have total control such as Codeigniter, Cakephp, etc.
Hope this helps.
Ben this is an interesting topic and no matter which direction you eventually go it would be nice to know your choice of combination. Depending on your requirements.
For something with medium range functionality OSCOMMERCE might be worth a check on. Also see Zencart, Virtumart and Megento.
I am creating the cms for a relatively simple site - portfolio, some general content pages, custom blog etc.
What are some of the best patterns to consider before diving into the design.
I want the system to be as flexible as possible without being too complex.
I have looked for some good resources that discus cms and blog design but can't find anything too good.
My language is php but I suppose I am looking for more language independent advice.
Flexibility without complexity... nice program.
Maybe you're a genius and you will make something that feet your needs. But I think the biggest problem you will face is security and robustness. So really, take other advices on this page and have a look at wordpress, drupal, joomla and ezpublish. A lot of security stuff is already done. And not only security...
So, study some of these tools, track their flaws, check their security policy. Study how they handle caching, sessions, bootstrap, absolute & relative url managment, documents (images, videos, etc), ajax, authentification, identification, acl, user interfaces, rich-text editing, migrations, templating, page composition, content filtering (I try to remove the things you won't need, plugins, database abstraction, fine caching, css and js minification, all the extra-complex stuff not needed for a single instance simple CMS). Soon you'll have a 'picture' of the stuff they've done.
By doing this work, you'll certainly notice some big differences, and mistakes. You'll start going on irc and flaming developpers, telling them that others have done better choices. You'll start forgetting to shave. You'll maybe do some contributions. Some will be accepted, others won't. Old core devs doesn't like when someone explain why they made mistakes (and they make mistakes).
Now, comes the day you have a beard. Some of your contributions will start looking like forks. You will have ennemies, and friends, or followers. And you will start feeling the force.
And you will go on irc and tell god that the world is ugly and that you'll make the first CMS which will be flexible without being complex. And people will cry. And birds will run in circles. And you will be able to explain what are the design pattern of a CMS.
I am a user. I know what I want. Doing what I want will make user happy. I'm happy.
You shall not trust code from people with glasses
"MVC MVC MVC" : and the people responds 'that shall be done'
Seriously, There's still a place for a good CMS with disruptive innovation, the fork history has started long time ago with phpNuke (as far as I can remember). But some of the actual products are really fine for most tasks.
I'm probably risking the reputation here, but my experience shows that building your own CMS can be a very justified decision, especially when you get familiar with current opensource systems and understand what exactly they lack in terms of features, security or what not. Open-source often means a lot of backward-compatibility concerns and bad architecture decisions that cannot be easily changed.
I strongly suggest that instead of just taking on MVC you take a look at ideas that make it attractive.
One main problem with CMSes is the range of technologies involved in driving dynamic web-sites: imperative php for logic, declarative SQL for data queries, markup HTML for interface, imperative/functional javascript for dynamic interface, JSON for ajax calls etc. To keep the system manageable you have to keep these technologies in a controlled and understandable environment, but yet allow for smooth integration. Knowledge and best practices are out there. MVC is but one approach to manage this problem.
My choice at the time was to use the following principles:
Object-oriented code with static calling (php is a one-run thing, many instances of code objects are rarely justified), nothing except for one line of init code in global context
100% code-design separation with the use of XSLT and custom content processor
Custom router that can take any http request and reroute it to registered methods
Custom content processor that can take arbitrary method output and convert it into any usable format such as xhtml, xml, json etc. based on the request parameters (i.e. http://local/class/method.xhtml, http://local/class/method.json)
One copy of code for as many virtual web servers as necessary
SQL query builder (chosen for flexibility over ORM) for all database queries
Mandatory filtering of method input with filter_* functions
I believe you can choose a few that you like :) And good luck!
A good pattern to start with is the Model View Controller pattern, or MVC.
This pattern suggests to seperate your application's logic in the following layers: data logic(model), manipulation or business logic (controller) and display logic (view).
This is a good pattern to start with as you'll run into other problems (and thus patterns) along the way.
The following website explains the MVC concept quite well: MVC Principles
There is no point reinventing the wheel unless you are trying to better it in anyway.
THere are a lot of CMS available already. I personally have worked with ezpublish. There are other options such as drupal etc. This is the list of all open source cms avaliable - Click here
If you are just trying to learn then you can perhaps pick any one of the popular opensource and work on them to find its architecture and design.
Besides, I dont think anyone can give you a list of design patterns that would be best for a CMS tool. Because each design pattern solves some particular problem. And, you just have to choose a design pattern depending on a specific problem you want to solve in your project.
These days, writing your own CMS is a horrible waste of time. The usual open source solutions -- these days Joomla, WordPress and Drupal are popular -- are written by thousands of people and while you might loose a little flexibility by using on that's ready made this is by far offset by not needing to redo everything from scratch. If you go with Drupal, you can also enjoy high quality, massively scalable etc code :)
If Your rquiremnt is portfolio, some general content pages, custom blog only, Wordpress will be simple and Better.
In PHP so many CMS available , most popular one is Joomla.
I'm starting to architect a quite complex web application. The implementation is probably going to be done in PHP, though if there are impressive reasons to choose a different environment I might be convinced.
I've looked at tools like Symfony and CakePHP. The problem is that it feels like they're relatively low level for a modern Web 2.0 application. They handle the basic things like MVC and scaffolding, but not the more advanced UI elements that I'm looking for. Here are some of my requirements:
Single page architecture. With minor exceptions, there should be no page refresh. All actions are done via ajax, the way it's done in gmail, and to a lesser extent in Facebook.
Ajax layout and widget handling. Not only the application doesn't refresh the page, but the developer can specify the layout and load various widgets into different parts of the page. This is somewhat like iGoogle, but should be better integrated.
Support both on the client side and server side for AJAX widgets. It should be trivial to display the result of a select statement in an AJAX table/array like http://developer.yahoo.com/yui/datatable/. This should also apply to other widgets including
Trees
Menus
Forms
Speaking of forms, there should be easy integration with client side validation
Signup/Authentication/Authorization. Including all the housekeeping things like forgot my password, CAPTCHAs, etc.
There's more, but I think I've given enough details so that you get an idea for what I'm looking for. Basically, I'd like to engineer a modern Web 2.0 app and skip writing, testing, debugging things that most web applications need to do. And yes, I know I can take YUI or jQuery and slap it on top of one of the regular platforms, but then I'd have to write all the glue. Now if there are modules that do this, that would be interesting.
So if you say, take Symfony + modules xyz + jQuery and there's your answer, I'd be happy to hear that.
Finally, in terms of priority, I'm looking for something that's scalable, reliable, well engineered more than something that's easy to learn and deploy.
Since you are looking for all things AJAX, why not try GWT? Its not PHP I agree, but it makes writing AJAX applications easy for developers.
I agree with #Iznogood. What your looking for is not actually a framework but a talent.You can make any framework as "igoogle-ish" as possible if you know what your doing.
I suggest look for a framework with a huge and active community like, CI, Cake and Jquery. Search/ask the community for the specific things you need. Plug it in and Presto!
But I'm afraid you'll have to write some of it.
If you don't mind using Java, there is ZK:
http://www.zkoss.org/
I'd prefer using this over GWT.
In php, you might want to consider Cjax http://cjax.sourceforge.net/.
It is MVC oriented, and has a very generic API, with full support for customization, including the development of plugins (There is a full Plugin API available, including documentation, Plugins can be built in PHP and JavaScript in combination -- see uploadify and validate plugins).
It can access all JavaScript functions, Objects and proprieties, and elements properties and functions from PHP server side, with one line of code.
In Cjax,
you can execute ajax actions, inside controllers (ajax requests) - without a line of JavaScript. this allows you to fully function without refreshing the page. It also allows you to access all Js objects from PHP. A good sample that reflects this is the "recursive ajax request" see that demo. The API can be used on page load, when the page first loads, and within every single ajax request.
Do take a look at the documentation and to the 20+ demos included, and no where you will see a single line of JavaScript. It allows you to manipulate elements, containers, request ajax, submit ajax forms, creates overlays, upload files through ajax, all from the server side. Take a glace at the API Table.
It has support for both, server side ajax, and client side, (see plugins JavaScript Documentation)
It plays nice with jQuery - the validation plugin in jquery is full executable in PHP without a line of JavaScript (see http://cjax.sourceforge.net/examples/plugin_validate.php without any custom line of Jquery inside the framework, it's all plugin's work). Ajax Responses from Jquery also get processed by this framework automatically, so using Jquery's Ajax function API wise, is the same as if you were using the Framework's Plugin's JavaScript Ajax Functions.
And you would just have to take your time learning more about it, because it has quite a few more tools that I am sure you'd find useful.
Currently there are two official releases, the generic release and
there is an official release for CodeIgnater of this framework (and it is the leading ajax option for codeignater) , and works great in conjunction other PHP Frameworks and without them.
Signup/Authentication/Authorization. Including all the housekeeping
things like forgot my password, CAPTCHAs, etc.
This is something that you can build within an ajax controller, the framework itself its meant to be a generic "AJAX" framework, so if you are looking for none ajax features, you will need to build them or integrate other PHP Framework with Cjax (such as Codeignater).
I do not know if this matches all of your criteria, or if you ever will find one. However, I like the Zend Framework myself.
As for the UI and AJAX portions - the server side technology matters little, and it's more about browser-side technology and interaction mechanisms, as well as DOM manipulation.
jQuery is my favorite for that. As for the PHP back-end, I tend to develop it with my own codebase that's grown with me over the years.
But all this takes a lot of practice, knowledge, education, research, reading, and posting questions on StackOverflow.com ;)
I would recommend symfony PHP framework since it has very good support for every feature you mentioned, and it can be easily integrated with Zend Framework - as of version 2.0 coming this year it would be integrated in the package. You can easily set symfony to handle AJAX requests so that it would be perfect transparent layer handling server side.
For me, if you look at THAT heavy AJAX, I would recommend searching for some good JavaScript framework - such as Ext.JS - that would handle all client side functionalities.
For the front-end part of your app, you might want to have a look at SproutCore ( http://www.sproutcore.com/ ).
Building a Web application with SproutCore feels more like configuring components than writing code.
There is no glue code.
On the backend all you have to do is accept and emit JSON.
Depends what sort of level you're looking at. If by 'framework' you mean something like Zend or Symphony, then to be honest all the big ones are about as good as each other. They all have strong points and weak points, but none will really meet your criteria.
But your question implies you're looking for something more than that kind of framework. Maybe you're looking for a full-blown CMS platform like Drupal, Joomla or Wordpress?
In that case, again, you need to consider your needs verses the capabilities and pitfalls of each system.
Drupal, for example, has masses of modules, is very powerful, and easy to write your own modules, but isn't object-oriented and doesn't really do MVC, so if that's your bag then you may find it hard to get on with. Wordpress is much easier to get going quickly but is less flexible once you start getting deeper into it.
But again, they all have strengths and weaknesses. If one was clearly the best, it would be an easy choice. But at the end of day perhaps it's better to have several good quality options to choose from.
Have a look at Ext JS, it is pretty good. If it is a commercial project you are doing it isn't free but not too expensive either. It also has a GWT version if you prefer that.
This is only for the front end but it is not too complicated to use any backend that can emit json.
CakePHP is pretty good
I understand the classical definition of a CMS: it's a "webapp" the main purpose of which is to handle "content", probably that's generated by its users (kind of like all of us here at SO and the content we provide is text and code).
I also always got the impression that creating a CMS is supposed to be a Really Tough Thing. But how so? Isn't a CMS just a webapp like any other. I would guess that many webapps are tougher to code than a traditional CMS.
If I were to think of creating a CMS like creating any other webapp, would I be wrong?
I suppose you can consider a CMS as any other kind of web application ; only thing is there are lots of functionalities that are both required, and/or expected by users.
A few of those which come to mind :
authentication / access control
to the application, of course
but also to every kind of data
and even, maybe, each field of each type of content
a bit of ergonomy -- especially if targeting not technical users.
dealing with media (photos, videos, music, flash content, ...) ; both upload, linking, and consultation
if designing a CMS that's not oriented toward a specific website, you have to create something generic, that can work even in cases you didn't think about
extensibility : you will create the core of your CMS ; but users will most likely want to add some additional functionalities
which requires a nice extension/plugin system
internationalization / localization
Well, that's just a really short list, actually ; and there are probably lots of other ideas that could come to mind...
So, yeah, a CMS is a web application -- but it has to be generic, if you want it to work for more than just one (kind of) website.
Read up on CMS and DMS in wikipedia. Then it should get clear that a good CMS is really a big task.
Especially templating, workflow (fixed ones, support for custom ones), user hierarchies and so on... are hard to get right and still make them easily adaptable and customizable.
And don't forget that a CMS doesn't have to be a WebApp, but there are so called WebCM-Systems.
Aside this WebCMS are like every other big complex WebApp.
Content management system
Document management system
I'm in the process of designing a PHP-based content management system for personal use and eventually to be distributed. I know there are a lot of CMS's already out there, but I really haven't found one that meets my all of my needs and I also would like to have the learning experience. Security is a large focus, as are extensibility and ease of use. For those of you out there who have built your own CMS, what advice can you offer? What features are essential for a core? What are must have add-ons? What did you wish you knew before starting? What's the biggest potential roadblock/problem? Any and all advice is welcome.
Edit: Any advice on marketing do's and don't's would also be appreciated.
In building a few iterations of CMSs, some of the key things turned out to be:
Having a good rich text editor - end-users really don't want to do HTML. Consensus seems to be that FCKEditor is the best - there have been a couple of questions on this here recently
Allowing people to add new pages and easily create a menu/tab structure or cross-link between pages
Determining how to fit content into a template and/or allowing users to develop the templates themselves
Figuring out how (and whether) to let people paste content from Microsoft Word - converting magic quotes, emdashes and the weirdish Wordish HTML
Including a spellchecking feature (though Firefox has something built-in and iespell may do the job for IE)
Some less critical but useful capabilities are:
- Ability to dynamically create readable and SEO-friendly URLs (the StackOverflow way is not bad)
- Ability to show earlier versions of content after it's modified
- Ability to have a sandbox for content to let it be proofread or checked before release
- Handling of multiple languages and non-English/non-ASCII characters
Well, building your own CMS actually implies that it is not an enterprise-level product. What this means is that you will not be able to actually implement all features that make CMS users happy. Not even most features. I want to clarify that by CMS I actually mean a platform for creating web applications or web sites, not a blogging platform or a scaled-down version. From personal experience I can tell you the things I want most in a CMS.
1. Extensible - provide a clean and robust API so that a programmer can do most things through code, instead of using the UI
2. Easy page creation and editing - use templates, have several URLs for a single page, provide options for URL rewriting
3. Make it component-based. Allow users to add custom functionality. Make it easy for someone to add his code to do something
4. Make it SEO-friendly. This includes metadata, again URL rewriting, good sitemap, etc.
Now there are these enterprise features that I also like, but i doubt you'll have the desire to dive into their implementation from the beginning. They include workflow (an approval process for content-creation, customizable), Built-in modules for common functionality (blogs, e-commerce, news), ability to write own modules, permissions for different users, built-in syndication, etc.
After all I speak from a developer's point of view and my opinion might not be mainstream, so you have to decide on your own in the end. Just as ahockley said - you have to know why you need to build your own CMS.
If you ask 100 different CMS users about the most important thing about their CMS, you'll probably get 80+ different answers.
The biggest roadblock is probably going to be people asking you why you built a new CMS from scratch.
If you don't know the answer to that question, I'm not sure why you're going down this path.
One thing to keep in mind is that for an internet CMS, folks are going to want integration points with many of the "usual" services. Leverage existing services such as photo sharing sites, Twitter, OpenID and the like before building your own proprietary solutions.
well i wrote a CMS for personal use and released it to the biggest chorus of chirping crickets ever! no biggie, though. i did learn a lot and i encourage you to move forward. my clients use it and like it and it's holding up fine.
but if i were to start over (and i might) here's the advice i would give myself:
scrub everything everything everything entered from the user
user administration is a product differentiator. bonus points for being able to handle someone copy/pasting from WORD.
extensibility. 90% of the comments i get are from developers who want to use the cms to host "some" of the website pages but not others. or they want to embed their custom scripts into the page among the content. my next cms will be as modular as i possibly can handle.
many folks are absolutely fanatic about clean urls.
From marketing point of view:
1) Make it templateable.
2) Make CMS SEF and have SEOed URLs.
If you need to build custom functionality where your CMS is really a window to the rest of your business layers, then use something like PyroCMS or FuelCMS which are based off of CodeIgniter framework.
Developers usually get lost in the weeds with Drupal and Joomla! / Wordpress quickly become spaghetti code-laced doozies over time. Its how much you have already drank from the Kool-aid punch bowl.
I know this isn't a direct answer to what you're looking for but if you haven't looked at it yet I'd recommend checking out CMS made simple. It has much less bloat than other CMS's and is fast and efficient. It's open source so it may be a good reference point for any questions you will run into.
Just use Drupal.
Out of the box it is very light and fast. You add modules for virtually everything, so that can be daunting but it is fantastic.
Its secure (NASA and The White House use it), its modular, its open-source, it is well supported, has a reputation for clean APIs, and has hundreds of modules from SEO to Wysiwyg....