File Footnote

The AllPairsExample introduces suplemental information in the form of footnotes that are added to the parsed html just like any other annotation. The footnotes appear as hyperlinks of the form [1], [2], [3] and so on.

Try http:run.cgi to see footnotes here.


We call these file footnotes because they link to supplemental files that we generate in the course of annotating a document. It is an idea that is simple in concept but tricky in execution. Consider the following.

  • legal/unique name generation
  • file creation permission
  • url construction from file name
  • document mobility
  • ambiguous lifetime (last reference)
  • accidental proliferation

We've made the following choices so that we can gain some experience with these mechanisms before trying something fancy.

We use simple integers as filenames, 1.html, 2.html and so on. We insure that they are uniquie only within a single run.

We store the generated files in a fixed place with the reletave address, "Results/footnotes". The RunScript must have sufficent permission to write in this directory.

We create relative links to the generated files by prepending "footnotes/" to the generated name. This presumes that the citing document will be found in Reports.

We added a symbolic link to the footnote directory from the wiki main directory. Without this, results from the RunScript,which appear to the browser to be located here, won't be able to find their footnotes.

We leave footnote files until they are overwritten by another run. The RunScript deletes reports after transmitting them back to the requestor. It doesn't delete related footnotes because it has no way of knowing if and when they will be retrieved.

We limit the number of footnotes to 25. Subsequent footnotes will appear as an unlinked null citation, a hyphen within brackets, [-].

See the source, String footnote (Parse tables) in this fixture.


Last edited November 5, 2002
Return to WelcomeVisitors