Error recovery parsing

The last weeks I have spent some time working (in fact, to be realistic, I was just scratching some problems) on the code folding plugin Coffee-Bytes for Eclipse. (I start loving the code folding idea a long time ago, since the first usage of JEdit - if I remember well, and for a long time I was wondering why so many editors do not use this feature. But about this maybe other time). The current investigations are for an error recovering parser. What I mean by this? A parser that is able to continue parsing (and maybe even return the broken nodes skeletons) after a failure (an unrecognized structure, a structure not according to the grammar). I have some backgrounds on writting parsers (using JavaCC and ANTLR), but till now I have written only parsers that abort the operation on the first error. Moreover, this time I have to rely the parsing on Eclipse support. This job seems pretty interesting, but as you probably know all the documentation I have is the javadocs. So I would like to hear from you any ideas how such a parsing must work.