Ionrock Dot Org

I started messing around a good deal with Plone. I haven't had a chance to actually develop anything for it just yet, but the reason is that when considering a CMS outside of the theme, Plone is very able and featureful. There are some problems and since Plone is a very mature project with a long history, there is an entire vocabulary related to it and its underlying technology Zope. I am hoping that because it is written in Python that the language can keep me engaged enough to work through the more difficult aspects.

Outside of Plone, I am planning on taking a look at Pylons. This is yet another python web framework that is really only exciting to me because of Paste. Paste is really interesting because it claims it is a framework for developing frameworks. As cheesy at that sounds, the actual concept is very solid. The idea is that libraries are rather helpful and often are used within projects without people rewriting massive amounts of code. Frameworks on the other hand consistently are rewritten and suffer from the NDH (Not Developed Here) mentality. This may be due to many programmer's desire to learn, although I think it merely supports that a developer is willing to use another's library because it doesn't make any suggestions to how the software should be designed. This is somewhat logical as software design decisions are always changed when things hit the real world. A framework solves a problem in a specific scenario, while a library is really just a tool or feature that you can add to your own design.

This seems problematic because this doesn't emphasize good software design. This is a problem, but the reality is that most developers will eventually believe a framework of some kind is a good idea. This may invariably change the way they think about software design (most often times in a positive light) and in turn, they will move on to write better designed systems using only helpful libraries.

The one issue that I have found to surface with most frameworks is the problem of distribution. Rails for example is really a great framework when the develop gets to distribute the results on a controlled server. This is well and good, but PHP has does well because every web host on the planet supports it exceptionally. Paste also makes an effort to address this through using Eggs and distutils, which is great because it is built into the language.

Paste also allows you to create aspect oriented features through middleware. Again, this is very useful because most people aspect oriented features as being something to be done in a language. The problem is by pushing aspect oriented features such as security or permissions into a different layer of some system stack, you start to gain abstraction and scalability while keeping an easy distribution medium. If you want an example of this, you are looking at one. The stack known as the Internet has been proven extremely well equipped to be flexible and robust while keeping simplicity. I think the Paste and the rest of the Python frameworks that are moving towards the kind of architecture that Paste suggests will be the frameworks or systems left standing when the hype of web frameworks dies down.