Glenn Marcy wrote: > Actually, you can set the "continue-after-fatal-error" feature with > any parser, not just XNI. However, keep in mind that the feature > is considered to be defective in its implementation. In particular, > the reason to continue processing after fatal errors is to allow > the parser to report additional errors beyond the first one, not to > continue getting normal document events, which is in violation of > the XML spec. That being the case, creating a dependency on the > defective behavior is not advised.
hi, unless we specify somewhere class of defects that could be recoverable i think that the parser behavior should stay undetermined. however in this particular case when parsing of this kind of _defective_ XML it is really nice that Xerces2 will continue parsing allowing for some kind of recovery. > However, I can see no reason that you could not create a handler > that processed your bulk documents in chunks, allowing it to find > any well-formedness errors in each chunk, and would only need to > go back to the beginning of that chunk to "reparse" the corrected > chunk before going on to the next one. how would you implement such chunk processor? i think that it would require to have a second lower level and more tolerant parser (or maybe just xml tokenizer) that would sift through data to find chunks that are then handed off to full XML parser? thanks, alek --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
