BlogNifty

Release 18 is the latest version of Cognifty. The new big feature is custom content publishers. You can create your own types of content based off of the main file upload tool or text tool. You can write your own plug-in to handle the publishing and do any sort of compiling, checking, or linking required before the content goes out to the site.

The blog system has been ported over to this custom content publishing framework. To make your own publisher you can use the blog publisher as a starting point. To activate your publisher simply add one configuration line to the default.ini

;This is a default content publisher which is required for the blog module to function
content.extrasubtype.blogentry=
    @sys.path@/modules/blog/lib/Blog_Content_Publisher.php:Cgn_Content_Publisher_Blog

The content.extrasubtype. is required, anything after that must be a unique identifier for your content type.

I have been using this to make a client support portal. I upload files and set which client organization that file belongs to, then when I publish it, I can move the content to whatever database table I want, or publish it to a disk cache.

Other ideas could be:

• Create a movie review content type which requires an IMDB link before publishing.
• Make an e-book content type which could convert wiki syntax into PDF.
• Split publish a text file into an article and a knowledge base entry.
• Schedule your publishing by saving the content to another system or another database table.
• Push your content to a static site build tool, like a specialized cruise control.
• Push your content to Subversion or other source code manager.

Released last week, Cognifty 17 brings a new identicon library as well as a simple way to backup your database.

Identicons are a nice way to visually identify something. Most of the time, people confine their use to blog posters who don't have a specific avatar image. Assuming that the poster's IP or e-mail address are unique, the identicon is a visual representation of that piece of data.

I wanted to do something a bit more useful for a site manager than simply identifying anonymous commenters on a blog. So I've added identicons to each piece of content in the admin system: specifically each piece of content which is not an image. Instead of a an icon which simply represents the mime type of the data, each page, download, article, or blog post has a unique image displayed next to it. The idea is that an editor can become more familiar with frequently linked content and identify the correct document much more easily when looking at hundreds of pages.

The identicon is based off of the content's GUID, so it will never change throughout the lifetime of the content.

The library supports two different painters, one for GD and one for SVG. The default "Geometry" based identicon can be extended for other graphic displays, such as words, fractals, or cartoons.

The standard way to create an icon is pretty simple:

 $seed = md5(rand());
 $size = 64; //pixels
 $blocks = 16; //complexity of the geometry
 
 $t['icon'] = new Cgn_Vis_Identicon_Geometry($seed, $size, $size, $blocks);
 $t['icon']->buildIcon();
 
 header('Content-type: image/png');
 echo $t['icon']->getIcon();

I didn't blog about release 15 specifically, so I'll combine the new features from 15 and 16 into one post.

Release 15

Release 16

In release 14 of Cognifty I cleaned up the CGI request handler to work with nanoweb. It's pretty different from the Apache CGI handler for PHP because it doesn't break your PATH INFO, etc. I've always been interested in running "self contained" PHP applications that can be distributed with their own server, at least as a way to demo a product. So, that interest has led me to live CDs, and Apache Friends' XAMPP.

Now, however, I'm all about the nanoweb. You can build a PHP-CGI binary for 1 platform, compile in PDO and SQLite (since sqlite is just a C client library and not a server), and ship it with nanoweb. Now you have a full Web application stack in only about 14 Mb (depending on what you throw into the php-cgi and how much you compile statically).

I think the real benefit of this setup is being able to run an app that is completely self-contained. It doesn't matter if someone has apache installed or not, it doesn't matter if they have php 4.3.0 with safe_mode on, it doesn't matter if they have mysql-5 server but have compiled PHP against mysql-4.1 client libraries. With this sort of setup, your app is completely segmented from what is on the target system.

So, what benefit does being completely segmented have? Well, look at Plesk. It's a complete virtual host manager distributed with it's own Apache, PHP, and Mysql so that it can stay running even if you break the system's default settings to the point where nothing works.

In honor of this release's compatibility with nanoweb and the new SQLite PDO driver, I've written the first part of an article dealing with how to write your own virtual host control panel, or other embedded PHP servers.

Read the article, leave comments here. Tell me if the code is working or not. All the code is available on sourceforge.net's Subversion here.

svn co http://niftyphp.svn.sourceforge.net/niftyphp/niftyserver/tags/tutorial-p1 niftyserver

I've just added a new "live cheat sheet" to the http://cognifty.com/index.php/tutorial/ section. I've seen some other projects which have these so-called "cheat sheets", but the cheat sheet is simply a huge PDF. These PDFs usually show a graphical representation of the software system as a set of colored blocks and arrows with tiny variables written inside the blocks. These variables are supposed to help you remember which functions are available at which points in the system.

While it's nice to have an overview of a system with colors and blocks, I don't really want to have to have a PDF reader loaded while I'm developing. So, I'm trying to do a sort of "live cheat sheet", in which you find what you want to do, you click the words, then you are able to simply copy and paste the sample code into your application.

I thought this would be good since a couple of people are always asking me how to do this and that in Cognifty. To me, it's very easy because I know exactly where each type of example code is located. I know that the blog system has the best implementation of the Spam blocking Trusted Service framework. I know that the admin system makes extensive use of the data grid MVC. The problem is, other people don't know these things off the tops of their collective heads.

I'll admit it, a fair amount of my coding time is spent using VI to copy (yank) and paste code from one module to another. I do this even if the code will require significant changes, because it seems a bit easier, or perhaps safer. But, how many other people copy all the boring "scaffolding" from one class to another? For frameworks that generate "scaffolding" code, I would consider typing the command to generate the code just as time consuming as copy, paste, search, replace.

I've already used this system myself to remember the exact class names and function names for some template methods. So, I believe it has some validity. Not only does it help jog a developer's memory and cut down on hunting for good copy/paste code, I think it gives new people a very good idea about what their work will look like inside a framework/application. Instead of seeing all around the code (the class definitions, the function names, the parameter signature) one can actually see inside the code, if you will. This type of code makes up the the day-to-day meat of coding in a system, not the elegant back-end or class hierarchy, which may or may not directly impact a developer.

Is this type of system obsoleted by IDEs with code completion? Is it useful? Does it give anyone a better insight into the CGN code-base?