CL-Webframework

Thursday, February 28, 2008, 09:54 PM - Computer, Common Lisp, Web Framework

I fetched up the idea of the CL-Webframework mentiond three postst before. I tried to implement an prototype in PHP because this server is a virtual machine and I have some problems to install SBCL.

But the Problem was that I have to deal quite fast with technical restrictions of PHP. That means that I have to think to much about the language itself and if you are a programming-rookie like me it's not so easy to keep track of all these language restrictions. In this way for me it is easier to write in lisp.

I have a little bit expirience with Edmund Weitz' hunchentoot HTTP-Server and I think I can do it easyer with such an setup.

I hope I'll can work a little bit on this small project the next time.

PS:
My concept of the Framework is not much more like the dispatcher of hunchentoot but I like some more functions which are more specific to web publishing which makes content-providing more flexible and easier to use.
I think the way I'd like to go is to provide an more abstract layer for the dispatcher of hunchentoot which maps the ideas and concepts of Publishing Documents in the Web to an Server.
| permalink

Idea for CL-Webframework

Saturday, February 2, 2008, 11:41 PM - Common Lisp, Web Framework

The idea behind this framework ist that all sites are made of any number of content blocks which belong to an URL.
So the content blocks are mapped to an URL
(URL (Content-I Content-II))

An Example:

Content Block I

-------

| abc |

-------



Content Block II

-------

| def |

-------

Block I and II are bound to serverrelativ URL /content
(:/content ("Content Block I" "Content Block II"))
http://www.martin-enders.de/content leads to a page like this:

/-page----------\
|               |
|   -------     |
|   | abc |     |
|   -------     |
|   -------     |
|   | def |     |
|   -------     |
|_ _ _ _ _ _ _ _|

Idea

The idea behind this framework is that you can map differnt content blocks to an url.

Problems / Questions

How should sites be treated when you use special functionality. Is in content blocks just HTMl allowed or should it be possible to integrate lisp into the pages?
The content Blocks are integrated in an layout so the HTML in the content block is just a part between the <body> and </body> tag. If you need an extra css file you have to be enabled to integrate an special tag to integrate additional information into the HTML head.

But all these workarounds lead to inconsistencies in the framework - it seems that it's not a good design now.

Conceptual Lisp Dispatcher:

(defvar *dispatch*)
(setf *dispatch* '(:/       ("index")
		   :/web    ("web")
		   :/css    ("css2" "css3")))

(defun get-content(content-block)
  content-block)

(defun render-content(content)
  (format t "~a" content))

(defun dispatcher(script-name)
  "This function takes a string as argument 
   and converts it into an keyword for 
   comparison with the url's in *dispatch*"
  (let* ((request (intern (string-upcase script-name) "KEYWORD"))
	 (content (getf *dispatch* request)))
    (dolist (content-block content)
      (render-content (get-content content-block)))
    request))

Conclusion

I'd like to have a small consistent framework for publishing in the web. But the current state of this idea has to many inconsistences.
| permalink