10 abnormalities in CodeIgniter that everyone should know

[eluser]Armorfist[/eluser]

- the lack of modularity (please make Zacharias Knudsen’s library either official or obsolete).
- the validation callbacks inside controllers rather than inside a third party class. Each time I have to copy/paste a callback method from one controller to another I cringe.

I second that.

[eluser]sikkle[/eluser]
Modularity is also a question of choice, i mean, you can need different way to add this option.

So i'll suggest you give a good look to matchbox and also HMVC stuff.



I totally agree with the callback stuff

see ya around.

[eluser]Buda[/eluser]
[quote author="sikkle" date="1205694420"]To keep it short, don’t blame Codeigniter to stay at php4[/quote]

I'm not blaming CodeIgniter, I think it's a great open source project that has helped thousands of developers make the big jump from procedural to OOP. I have a big respect for this project and community. At the end of the day, CI it's a simple solution to a complex problem, and I'm glad that the solution was made open source.

[quote author="sikkle" date="1205694420"]Modularity is also a question of choice[/quote]

So is OOP, you don't know you need it until you need it. For small applications, you might not need objects, modularity or even an MVC architecture, but, what makes you so sure that the application is not going to scale or the requirements are not going to change?

[quote author="CodeIgniter" date="1205635902"]Your analysis seems based on first hand experience with other projects that have an architecture closer to your ideal. I'm curious as to which framework you would consider to build on top of if you were given the choice.[/quote]

Well, I'm a big fan of Django for example. And in PHP I like what they did with Zend_Controller. But each framework has its pros and cons. I'm quite surprised what CakePHP managed to pull out, last year, when everyone thought that its time had come, the community became more proactive, refactored a lot of the code, started promoting it, and are now stepping Zend's tails. And no one saw it coming.

I think that's what this community needs, at the moment this community is more reactive that proactive, waiting for something to happen. Cake is a good example that an open source project can re-invent itself. Maybe CI can do the same? At the end of the day, I think it's all up to the community.

[eluser]Crimp[/eluser]
Good posts.

This may not be what you want, but I usually extend CI_Validation to include common validation methods in a project: MY_Validation.

[eluser]xwero[/eluser]
@CodeIginter : Modularity is not always needed. I don't think CI should enforce it.

[eluser]Michael Ekoka[/eluser]
@xwero: Don't forget the days when you were a complete noob and had no clue about architecture. Remember that you also, were once introduced to a lot of really cool concepts that you had no idea about by the means of an official distribution and/or document and even though you don't use them all, you're grateful to be aware of them and that they're available.

Matchbox is Zacharias Knudsen's library, but it's not an official library, nor does it get a mention in any official CI document. When they upgrade CI and change the core Router.php and Loader.php files, they don't stop and think "hey, what about Zacharias's library?" A CI user could easily code for a year without ever hearing about Matchbox. It's not ok to have to look for a concept like modularity in order to find it.

The purpose of an open source project is not only to have something available for free for people to use, it's also about community contributions and growth. For one, having support for modules doesn't mean that one has to use it. Second, a lot of things in CI are not always needed, but they still find their way into the framework. Modularity is an invaluable feature when you want to build applications that involve an undisclosed amount of contributions.

It would be nice for example to have a repository of modules that I know I can just drop in my install and use without having to tweak files left and right to resolve conflicts that stems from multiple extended core libraries.

There are lots of people out there asking "why no cms? where cms? cms please?". Well, a modular CI would probably solve 90% of their problem. Do you need an User Account Manager? here's a module for ya. How about a Translation module to manage your multilingual contents? Hey you know what I think I also have a Picture Gallery manager somewhere here. A CMS would then become nothing more than a collection of modules with a common visual theme.

Module support should be made official because it makes sense. What would be frustrating down the line would be to have 3 or 4 solutions out there around which people are building libraries. Or to have a good chunk of contributions centralized around, say, Matchbox over an extended period of time, just to have CI finally releasing module support in version 2.8, that is not compatible with Matchbox.

Buda 16 March 2008 01:10 PM
at the moment this community is more reactive that proactive, waiting for something to happen.

I don't think the community has much choice. The infrastructures present for CI allow us to do exactly what we're doing now. Forums, Wiki, Feature Requests and Bug reports. Some of us go the extra mile and log on to irc or bookmark each other's blog articles. But there is not much more than that. I think there still is much to be done to really leverage what this community has to offer.

[eluser]dark_lord[/eluser]
We all have our own perspectives and judgment. I think one must used a framework which I think he/she is more comfortable working with. Everybody has its own logic, and that the deviation between programmers. They have their own way of finding a solutions to things they do and which would make their lives simplier. One may have a long method, the other may have a shorter one. Just code in a framework where you are comfortable with, and enjoy the work.

[eluser]sikkle[/eluser]
buda ;

i think this is exactly the point, pros and cons, usually have to be fair en show them in the same thread

thanks for your objectivity

[eluser]webthink[/eluser]
With respect to wish_bear... The thread is not really about coding with what you're comfortable with or enjoying the work nor is it about individual programming styles. It's about identifying some fairly serious shortcomings and potential performance/scalability issues . I've not had the time to have a look under the hood to see if CI is as effective as it could be so I'm glad that other people are taking the time bring any issues to light and discuss them in a serious manner.

[eluser]xwero[/eluser]
@codeigniter : I agree with you there should be a better way to deal with contributions than to put them on the forum or in the wiki. I've seen quite a few people use google code to store their contributions to CI. I have not a lot experience with google code but maybe it would be a good idea to create a group there and provide a search page on the code igniter site. They could even adopt a pastebin site to do the same for snippets. And anything else they can think of. This manner of working has 2 benefits : Ellislab doesn't have to invest in servers for third party contributions and people searching google code can become aware of the existence of CI through a third party library.

But to come back on topic i don't think the Ellislab developers shouldn't invest their time in making the framework bigger but making the framework better. This thread is a good example of the things that can made better in CI. Buda talks about the core of the framework.