Memory leak with #to_plain_text

[mksm]

I'm parsing a lot of very similar HTML pages. Here's a bit of code:

h = Hpricot(html)
table_fields = []
h.traverse_element('td') do |e|
  table_fields << e.to_plain_text.strip.squeeze(" ")
end

Tested with Hpricot 0.6, Linux, Ruby 1.8.6 (2007-03-13 patchlevel 0).

I've switched to #innerHTML and it works fine. Though it doesnt produce the same results.

Regards, mksm