Use the Find Page to search for documents with the given text in their titles, or for documents containing a given string of text. A "Full Search" of all text in all documents is the second option on that page.
[Fullsearch]
As suggested by Dave Smith, the full search examines an abbreviated index of all wiki content. This has the advantage of being much faster than searching the actual content and doesn't lock out updates while the search is in progress. There are, however, a couple of limitations...
The index can be up to a day old. You won't find recently entered text.
It indexes words only. You can't search for phrases or punctuation.
There are actually two versions of the index, one with all words and a second with Wiki Names only. The second index is much shorter and can be searched in a second or so. (The choice of index is automatic.) This optimization makes the reverse page references (the page title hyperlink) useful again.
You can also append a search string to c2.com 
.
The full search finds a page whose full name matches (non-case-sensitively) the search string, but only if the page currently exists; to return partial matches, the title search must be used. Also, a page name is found within a page only if searched for using a correctly capitalized search string. These facts help explain why Wiki's Reverse Index search works as it does. Some other wikis do things differently and use a separately-coded search for Reverse Index.
To trick indexing robots into thinking there is only one search results page, the search URL c2.com under the reverse link is the same for all pages.
But how does Full Search know which page to search for then? When fetching a new page, the web browser reports the URL of the referring page, and Full Search extracts the wiki page title from this URL. If you have disabled this feature in your browser, the search won't work. Since the URL does not change, some browsers will cache the result and hand you the same page every time. In this case, just hit refresh (reload).
Search Word Must Be Alphabetic is sometimes a hindrance (viz. xp2001 or non-Ascii Code like naïve). Would it make sense to allow requesting meaningful tokens to index?
Solve the Google Hates Wiki problem, you will have infinitely better search functionalities, and more participation in generating contents.
I am trying to find pages relevant to Train Wreck-style coding (e.g., foo.bar().baz().getBlort().withHonkey().kissTheDonkey(Body.getAnatomyFactory("lips").instantiate().onThe())). I can't seem to use the full-search to find the pages, and yet, I know that the phrase exists here, because C2 is where I first heard the phrase! -- Samuel Falvo Polymorphism Limits includes "train-wreck".
See original on c2.com