Excellent module, much easier than Template::Extract and HTML::TreeBuilder for extracting data from web pages in many cases, and one even doesn't have to look into the HTML source being processed.
My only complaint is the encoding problem. When dealing with pages in non-ascii and non-utf8 encodings like GB2312, it just refuses to match headers. I have to convert the HTML input to UTF-8 manually all the time. I think it may be a problem on the HTML::Parser side... So UTF-8 is always my best friend. :)
This module is extremely easy for both web site automation and web site testing. The Selenium server can control my Firefox 1.5 browser to access many real-world web sites flawlessly. The Selenium firefox extention is also very handy for generating the perl code (making use of Test::WWW::Selenium) automatically for me.
I've found this module the best choice for testing complex data structures like ASTs. When a test fails, it provides much more detailed info on how two deep structures differ than Test::More's eq_array and eq_hash do. The brilliant idea of this module reminds me of a funny language named TXL.
Testing ASTs could hardly be interesting without Test::Deep! So I decide to rate it five stars.
Yeah, the capabilities are currently very limited. I can't even figure out how to build an untrivial Visio page according to the current POD documentation. Anyway, automating Visio is really a great idea, so I wish the author to improve this stuff constantly.