Hello, maintainers:
For the past year, I've been developing a GPL-ed framework out of my code libraries which allows for rapid development of perl-driven web sites on Apache in an object-oriented way. I did this since Mason and PHP were too restrictive for my needs. I find it useful enough that I'd like to make it available to other perl users as soon as it's stable enough to be post-able and I've written up the documentation.
At the core is a handler, a sort of data-storage object, and a base object which parses the HTML text it is handed, recursively spawning new objects as they are represented in HTML. For example, the HTML snippet:
<Wyrd::Lattice>
<Wyrd::Query type="cgi">
select b.title as title, b.description as description
from authors a, books b
where a.id=b.author_id and a.id=$:author_id;
</Wyrd::Query>
<Wyrd::Lattice::Header>
<tr><th>Books by
<Wyrd::Lookup type="cgi" query="select name from authors where id=$:author_id"><b>$:name</b></Wyrd::Lookup>
</th></tr>
</Wyrd::Lattice::Header>
<Wyrd::Lattice::Grid>
<tr><td>$:title?:description{-- $:description}
</Wyrd::Lattice::Grid>
</Wyrd::Lattice>
Will extract an author id from the CGI environment (via Apache::Request) and use it to perform two queries, returning an HTML table with the phrase "Books by <said author>", their titles, and, if provided, the description. This table will be embedded at this point on the HTML page returned to the browser.
Also featured in this framework are conditional phrases (like the ?:description element above), a DBI hook for DBA interface, an encrypted cookie-authentication module, a search engine indexing engine, a search engine, a very robust multiple-page forms processing object, intelligent forms elements (inputs, etc.), and so on.
The beauty of it is that for different virtual sites, a different set of derived classes can be used, so that Wyrd::Lattice (a table creation object) becomes a NewSite::Lattice, different only in some hook-in methods as to how it is formatted in HTML. This way the same functionality is preserved, but with whatever tweaks are appropriate to the virtual server's site.
The ugliness of it is that it is heavily interdependent. It comes as a lump without a great deal of independent usefulness. For example, the Wyrd::Interfaces::GetUser class depends on the Wyrd::Handler class and the Wyrd::DBL class for information on the logged-in user, and the Wyrd::Services::Auth and Wyrd::Services::TicketPad class to handle the pre-access authentication. You can derive a new Auth object to handle authentication in a slightly different way for example, but the GetUser object will still expect all of it's methods to be available. Yuk.
Additionally, there is the ugliness of derived classes which mimic the Wyrd framework (e.g. Newsite::XXXX objects) pollute the local namespace of the server. In the intended application, this is not a problem since it's the sysadmin's duty to control the local namespace, but if new frameworks are derived from the Wyrd framework for distribution elsewhere, they might pollute CPAN namespace. Double yuk.
So, is CPAN an appropriate place to distribute this framework? If so, I would like to do so under the namespace WO or Wyrd.
Barry King Wyrdwright, Inc.
