RE: Sorting of DOM input to c14n 2.0

> One of the important goals of C14N 2.0 is performance, even in DOM. And
for
> that I made sure that we don't traverse any of the nodes that we don't
have
> to. Say you are only signing 10% of the document, it would be very
> inefficient if you need to traverse the whole document just to sign a
small
> part of it.

Sure. But some use cases aren't like that and could be generally as
efficient by avoiding the sort. I'm just suggesting that the processing
rules be written without regard for *how* the order is determined, merely
that it is. The pseudocode can include the explicit note about it (possibly
including the algorithm).

Thanks for outlining the algorithm. On the call I thought you were talking
about something more efficient or simpler than that one (meaning you didn't
have to traverse back down the tree after you look for the ancestors).

-- Scott

Received on Tuesday, 13 April 2010 17:56:54 UTC