usable in any place a human can be used


orm and sql

I've been working on a side project in earnest lately. It's all kinds of PHP fun and I'm enjoying learning the ins and outs of PHP 5.3 as well as relearning some of the stuff I already knew about from my PHP glory days. I've been looking at various different ORM solutions to use with my project and I'd like to take some time to review them, explain why I chose none of them, and what I'm doing instead. For the non-technical in the audience, ORM stands for Object-Relational Mapper, its a piece of software that allows you to save parts of your program to and load them back from a database, supposedly quickly and easily.

  • Doctrine
    Doctrine is the 800-pound gorilla of PHP ORM solutions, it has it all, and then some. It is an ORM sitting on top of a DBAL (Database Abstraction Layer) which leverages its own query language DQL (Doctrine Query Language). It can be configured in any number of ways, supports all kinds of backends, is mature, stable, and feature rich. That's all the good of Doctrine, the bad is the learning curve. The manual for Doctrine is 30 sections long, each section is quite a bit to take in. This is great if you are doing an enterprise level program, but for my project Doctrine was overkill.
  • Flourish Lib
    This is not an ORM solution, although it does contain one. Flourish is an unframework, and a really, really good one at that. If you want to shut someone up who says you can't write good code in PHP, send them to Flourish, the creator Will Bond did a tremendous job with this unframework, and I still plan on using large parts of it. The ORM layer is actually really nice, there is a bit of a learning curve, and at the end of the day I decided that it did too much and polluted my models too much. Flourish though is definitely worth learning, the website also has great best practices to follow if you are building a PHP Web Application.
  • php.activerecord
    Based off of the widely successful Ruby on Rails ActiveRecord class, this project aims to bring the ease of Rails database interactions to PHP. It does not attempt to be a PHP on Rails framework, there are plenty of those, its just a great implementation of the ActiveRecord pattern. The documentation is also fantastic, covering the essentials and letting you jump right in, it feels like there is no learning curve at all.
  • RedBean
    This is a complete departure from normal ORM solutions. In a normal solution you are cognizant of both the object model and the relational model, the ORM acts as a pleasant interface for interactions. In RedBean you are freed from having to know about the relational model, in fact you are allowed to let the relational model change on the fly. Need a new attribute for that object, don't worry about migrating schemas, just slap it in there and let RedBean figure out the rest. It is definitely an interesting idea, and it is maturing quickly, but I was wary of using it because of the overly fluid nature and the business constraints of my project

Those are the most interesting ones I investigated, I investigated quite a few other ones, but these were definitely the cream of the crop. None of them fit my project, but my project is a little bit weird (if you keep reading this blog you will no doubt see it one day, unless something shiny grabs my attention and I wonder off). If you are looking for a really powerful ORM with all the bells and whistles, check out Doctrine. If you are familiar with ActiveRecord, php.activerecord is a fantastic implementation. If you are programming PHP at all, take the time to read through Will Bond's amazing Flourish Lib. If you need some lightweight persistence or want to dabble in some object-oriented databases, give RedBean a try. Really on that last one, if you are even at all interested by technology, check out RedBean, it is a little young but shows amazing potential and is a great example of thinking outside the box.

So what did I decide, well I decided I don't want to use an ORM layer. ORM didn't fit my use case, I was only experiencing developer pain trying to shoehorn it in there. I decided that what I needed was something a little different, and I'm currently developing exactly that. So what is this mystery project that I'm working on, its a couple different parts that work together, but I decided that all I really wanted was the following list of things.

  • Cross platform SQL
  • Automatic CRUD
  • Lightweight library

So I'm writing them, and I will be releasing at least part of it soon, once I get it to a point where it does something, then expect a blog post with trumpets and whatnot. I conceived the project structure last night and began coding, I was able to put in 3 hours of work and got a very nice proof of concept running, but it is still all sharp edges and scuffed surfaces.

Stay tuned though, I hope to have something people can put their fingers on soon. I think there is a need for the lightweight components that I'm building, as a platform for future innovation and because after 3 months of looking around I couldn't find anything out there that did what I needed.

No comments:

Post a Comment