From: Lars Christensen <larsch@...>
Date: 2001-09-18T23:09:02+09:00
Subject: [ruby-talk:21323] Re: Suggestions for new CGI lib

On Tue, 18 Sep 2001, Anders Johannsen wrote:

> Since it is a model, actual functionality is yet to be implemented. I
> would appreciate a discussion of pros and cons in my suggestion.
> Comments and improvements are most welcome.

I think this looks good. Writing short CGI scripts will become a bit more
messy, but this may be required to build a solid framework.

Perhaps CGI::Request/Response::Cookie and CGI::Request/Response::Header
should be moved to HTTP::Cookie and HTTP::Header and each have the ability
to parse and format HTTP cookies and headers.

Perhaps you should also change $defout somewhere in order to intercept
normal puts command, so you don't have to write directly to the cgi
object.

> The goals of design are to:
>
> * Introduce the well-known 'Request/Response' model
> * Separate Cookie, Post, Get and Environment namespaces
> * Add support for 'plug-in' output handlers, thus making for example
> gz compression of output easy.
> * Use an open model to make integration with fastcgi/cgisock and
> future technologies seamless
> * Separate actual CGI and HTML generation, as they are two very
> different matters (my suggestion only cover CGI)
> -- while still maintaining 'ease-of-use'

-- 
Lars Christensen, larsch@cs.auc.dk