Hi monks,

I am planning the migration from HTML::Mason (version 1) to Mason version 2. It is a website of approx. 1000 pages most of which contain static html only. Mason 1 is very good at wrapping the contents of the files in a frame providing navigation and the usual stuff. Some pages have got additional intelligence that is realized with plain perl.

All pages have got a page title set:

<%ATTR>
title => "My humble web page"
</%ATTR>
[download]

You might ask why it's not simply static <h1>-Tags. But I realized the page title as an attribute due to the fact that this information, among other, is needed for our institutional web management system.

Now, the equivalent Mason 2 code would be:

<%CLASS>
has title => (is => "ro", default => "My humble web page", init_arg =>
+ undef);
</%CLASS>
[download]

'ro' means read-only, init_arg => undef makes sure the title cannot be reset to some 'title' param passed by URL query part. However, I see two problems with this approach:

• The need to touch and update all those pages. To be safe, this would probably be done manually.
• The pages are maintained by our web editors who are not authorized to touch or change code, yet to encounter as little as possible of what resides of it in the pages, but in fact <%CLASS> blocks contain plain perl, therefore they are potentially unsafe.

Is there something I missed in the documentation? Already considered subclassing Mason::Compilation, overwriting its unnamed_block_types to have it cover <%attr>, but then I have not the slightest clue of how to associate any handling code with it, and the code jungle in that module is very impressive, leaves me reluctant to duplicate and change any piece of it.

Unfortunately, CMS are not considered an option for us. Only Mason provides for enough flexibility regarding our needs.

Oh dear monks, I hope you can help me here.

Regards,
flowdy