I loved the warm and intimate voice of the cello, and I was amazed at the rich sound of such a small ensemble.

Overall, it wasn’t your average concert, not even your average “authentic performance” concert. A very refreshing experience — and excellent music, which is what matters the most to me!

Gardiner is far from being ready for public consumption, but it could use some scrutiny for interested eyes. If you want to have a look, you will need bazaar 1.6.1 or later to checkout the code, a recent PHP 5.3 snapshot to run the code, and PHPUnit 3.3.4 to run the tests.

Go grab the code with bzr branch lp:gardiner!

As this move was all I was waiting to upgrade Gardiner to the new namespace syntax, I took the plunge, grabbed the latest PHP 5.3 snapshot, compiled and… suddenly all of my code was looking like gibberish to the interpreter!

Fixing it all was not a matter of find-and-replace alone, but it didn’t take long and was not especially painful. The feeling was funny, though… All of your good code turning into a burning pile of syntax errors is striking. And the corrected code is still looking a bit silly to my eyes. :-)

Now, back to our regular developing.

(Martin Fowler, Refactoring. Improving the Design of Existing Code)

The code is also getting in shape for hackers to have a look at it (nothing that you can use right away, I’m sorry). I’m going to publish the trunk by November 15, and I might publish an experimental branch very soon. More details on this experimental branch, and the very first code samples, to follow soon.

(Andrew Hunt, David Thomas, Pragmatic Unit Testing In Java with JUnit)

In the process of reviewing the various licenses, I discovered something that everybody already agrees on: there are way too many free/open source licenses around, and this creates complexities for everyone. My main guidance in the process was chapter 9 of Karl Fogel’s excellent book Producing Open Source Software. David Wheeler’s article Make Your Open Source Software GPL-Compatible. Or Else is also filled with interesting information and references.

At the end of the day, the tool that helped me turn whatever little knowledge I had absorbed into something actionable was John Cowan’s license selection wizard. If license options are giving you the headache, just use it!

What I’m up to these days is, basically, sorting out things and taking decisions. None of them is code related, but all of them are required to go public. The primary concern is of course which license to pick; the second, but only by a small mark, is deciding what documentation to write for the go live, the main contenders being embedded code documentation vs. a short overview document.

Other practical concerns, like most practical concerns, sorted themselves out on their own. So, if you’re interested, or even just curious, go download the latest bazaar or go get a Launchpad account! :-)

As soon as I settle for one license and I write some more documentation, I will publish a branch with some preliminary work, including the foundation of the metadata and domain layer, and a small bonus in the form of a template engine (something I started as a diversion from object/relational mapping design, and I’m now polishing as a diversion from the management issues related above — so, there must be a pattern here!). From there, we can build up to the first public release proper, with a focus on the metadata-based O/R mapper.

Talking feaures, what I want to be in 0.1 is support for Entity as well as Value objects (with Dependant Mapping for Values), with support for INSERT/UPDATE/DELETE and some basic SELECTs, possibly with support for relations between Entities. Likely with support for just one database engine, either MySQL or SQLite. The template engine will of course be in, although not feature complete.

The notable absent from release 0.1 will be anything else web-application oriented, including the basic View and Controller infrastructure. That will be the focus for our second iteration… Oops, this post’s about 0.1, so I shouldn’t be talking about that yet! :-)

« To build a web application framework that enables domain-driven design in PHP through rich in-code metadata, dependency injection and Repository-based persistence. »

I also took some time to list the major features that need to be completed before I can call a 1.0 release:

• Domain-driven design support for Entity, Value and Service objects described by in-code metadata
• Repository-style, metadata-based object/relational mapper
  • can persist inheritance hierarchies and object relations
  • persists Value objects through Dependent Mapping
  • supports MySQL, PostgreSQL and SQLite via PDO
• Metadata-based dependency injection and management of Service objects
• Lightweight template engine, with no PHP code in templates
  • supports layouts and overlays for page structure
  • extendable through custom tags
• Clean, search-engine friendly URLs via custom routing rules

I know the details are going to change, but it’s good to have a goal and a direction. Later on I’ll report on the current status of things, or how far we are from that goal…

(K. Pugh, Interface-Oriented Design)

Just at that time, I had come to the conclusion that I needed a real name for my own project and was struggling to make up something good enough. Of course, the HOWTO not only has naming advice, it also refers to Leslie Orchard’s short but thought-provoking article On Naming an Open Source Project. The lesson in there is that you need to pick a name that is both informative and memorable, and that you’d better not be too clever about the name, if this means you pick something obscure or complicated.

Repeat it once more, and it will look so obvious that it goes without saying. But truth is I was verging towards the obscure and the complicated, and Orchard’s own project’s name, laijutsu, was such (to the point that even the HOWTO gets it wrong in the first place).

So, I started looking for something based upon the prominent traits of my project. It’s obviously got all of the usual suspects: it’s full-stack, it’s MVC, it’s meant for web application development, it’s got an orm layer. But above all, it’s meant to fluently support a rich Model, so as to enable true Domain-Driven Design in PHP development.

DDD is key, and I tried really hard to come with some clever, but not too much, combination of these words. And I failed. Spectacularly.

Suffice it to say that you risked being inflicted names like d³, or ddwd (domain-driven web development, to be pronounced dude).

It was at that point that I recalled that Django got its name from gypsy jazz guitarist Django Reinhardt. Because Django is influential to me, it suddenly dawned on me that I could pay a tribute to both Django and my own favored musician by calling my project after English conductor sir John Eliot Gardiner.

The Gardiner framework, or just Gardiner, sounds simple and elegant enough to me. Of course, it totally misses the informative part, but so does Django.

I think I can go with this… and I also fear I need a tagline as clever as theirs!

For now, let’s just set a mission for this blog.

My posts will be mainly about a project related to my professional interests, that is, developing a new PHP framework for web applications. Yet another one? you might ask. The short answer is, yes. The long answer will unfold as this blog goes on…

While we’re at that, from time to time I may throw in some fragments of thoughts and records from my other interests and activities, just to avoid monotony.

So again, take a seat and take a drink. I hope you enjoy the conversation as much as I do.