As far as I'm aware, search engines (on the web) typically index only the "user facing" content of a web page.
But what if you wanted to find pages(s) that contain a certain fragment of HTML or JavaScript. Is this possible?
Do any web-wide search engines provide this feature? Or do you think that it would not be worth implementing?
Yes, see Google Code Search and Koders. There are probably others. The Google page does seem to index code that appears on sites, try the lang:html & lang:javascript commands.
[Correction] It doesn't look like these sites index code within a site's source, only what is displayed on pages.
Not sure about the JavaScript, but I'm reasonably certain about any content in the HTML. I do some work with an open-source forum package, and while the version number is in a tool-tip, you can Google for the version number and see everyone who's using it (that's been indexed, natch.)
Opera has built a search engine for web page code, although it's not public yet called Metadata Analysis and Mining Application or MAMA, but they do have some interesting statistics/findings.