Honouring the IT tradition to eat one's own dog food, I have reimplemented my Google Cache experiment in Parrot assembly language, using the Budgie framework. More importantly, this gives me an excuse to blog about this again, which has become necessary as Google steadfastly refuses to spider anything except the front page. Specifically, I would like to get GoogleBot to also visit the three pages I have set up for The T-Files, Polly, and Living in the Google Cache to monitor how people find these sites on Google (note that these three links do not work yet, we first need Google to copy them into its cache).

I finally managed to install mod_parrot (and Apache2, on which it depends). So the latest version of my Polly example is now being served not as a CGI script, but as an Apache handler written in Parrot IMC.

I have this plan to supplement the Budgie templating system with a small web application framework (BudgieWebApp), that would hide the fact whether the code is executed as CGI or using mod_parrot. The initial version of BudgieWebApp can only show a single page, and there is no support for application logic or filling the page template with data, but it can already rely on some nifty features of the templating engine itself, such as conditional output depending on query parameters, or dynamic server-side includes.

With the amazingly fast progress of the wonderful Pugs project (just two months old, Pugs can already run the over twenty CPAN modules that have been ported to Perl 6, and there is an experimental compiler now, that compiles the Mandelbrot example into Parrot bytecode which, even though still totally unoptimised, runs at the same speed as the equivalent Perl 5 code) I am increasingly unsure if writing Parrot assembler by hand makes any sense at all, but it is nice to see how things come together.