Cart
[Home] [By Thread] [By Date] [Recent Entries]
Fuchs, Matthew wrote:
>
> I mentioned Murata Makoto's brilliant work with forest-automata as an
> expanded model for markup is that to gain the full expressive power of his
> work one needs to accept (in the worst case) bottom-up parsing (eliminates
> stream-based applications) and either exponential-time preprocessing (merely
> processing an unknown schema may be prohibitively expensive) or
> exponential-time validation (there are some schemas for which parsing is
> effectively impossible). Given that I can send you something which may take
> exponential time to process, schemas or instances can be used for
> denial-of-service attacks.
I don't believe this is the case.
A TREX schema can be processed in linear time (no preprocessing
to speak of other than parsing the schema) and can validate instances
with a stream-based algorithm taking O(n) amortized time.
I am not positive, but I think that the same is possible
for RELAX and RELAX-NG.
The trick is to build the DFA lazily. IMO, this is actually
easier to implement than the conventional approach that builds
an {N|D}FA as a preprocessing step.
--Joe English
jenglish@f...
|
