I just started a new job at a Louisville Web Design firm. They build websites, mobile applications and perform search engine optimization (SEO) for companies in the Louisville area.

I will be doing custom development for them. Now that I have a few weeks under my belt for them, I should have some more free time at home to continue working on Hardware Armory.

Bryan and I have been very busy rebuilding the back-end of HardwareArmory to be more scalable. Let me introduce you to a few of my favorite new features:

1. Our application now serializes the objects to XML instead of doing binary serialization. This means clients for alternate OS are now easier to develop.

2. We have scraped SqlMembershipProvider (I know, I know...I will explain below*).

3. I am in the middle of re-working our product mapping engine and product display engine.

4. We have now reached out to a designer!! No more giant, bubbly icons!

* As it turns out, SqlMembershipProvider and WCF from remote applications are just not ready yet. It was super easy to implement system-system internally, but as soon as you start distribution, it becomes a pain in the butt. We had a working solution, so we are going to revert back until the technology catches up without hacking the Gibson.

I am planning to take some time off work so that I can work on the site more consistantly.

I have been wanting to play around with Facebook as a platform for some time now and have finally gotten around to downloading an SDK.

The entire process from start to finish will be documented here on the blog. That should keep me busy for some time, and (hopefully) by the end of my postings there will be a working Facebook app ;)

I received some good advice (indirectly) at work today. The general message was to release early and release often. Don't bother nailing down a full set of features. Instead, focus on getting the basics of a few features out there and then let the user pool decide what works and what doesn't.

I am the only one working on the site at this point. I need to stay focused on getting the damn thing launched.

I have the need to pick up the MVP pattern for my job, so that means a lot of the new HardwawreArmory features are going to be designed using that pattern.

I can only think of one task, currently, that will be "Hello World" easy and that will be the creation of new computers (and accessories) since it is mostly data->form->data and doesn't require too much fanciness.

The only challenge I see is that the IHardware interface that all of the Hardware components use as a contract are going not going to be easy to integrate as they don't really work as a View the way they are designed.

There are a few benefits to this pattern:

1. Test Driven Development makes more sense. No code is written until I have created my tests. I could do this now, but it doesn't fit my current development flow.
2. Seriously separated concerns. UI is nothing but UI (Model). View is nothing but functionality (View). And the Presenter is the object handed between the two (Presenter).
3. I will learn this pattern which should be a good feeder into MCV (or just upgrading to a framework designed to assist MVP like Castle).

I think that I am going to start to put code snippets in here.

I worked all night long and was finally able to get the site up and running with SqlMembershipProvider instead of my home-grown solution.

Why?

1. It integrates really well with a lot of available applications written in .Net.
2. It is a well-received standard.
3. The heavy lifting is handled by the time-tested framework.
4. It is just THAT MUCH less code that I have to maintain.
5. I really like the administration (roles, specifically) interface that it gives me.

Now that I am starting to expose my WCF services to the world, I wanted to make sure that my login interface uses a standard of some kind. I will probably also upgrade to OpenID or something similar in the very near future as a secondary option to creating an account on my page.

Another cool benefit to my using SqlMembershipProvider is that it gave me a chance to re-evaluate how I was handling accounts in general. Previously, I was requiring lots of information to create an account. I simplified that down to simply take the email, username, password. I may have to integrate a CAPTCHA once people really start using the site to keep things spam free, but am at a stopping point for the login engine as of now.