Hi,
> Yes. The difference is that (a) the DOM-builder would have to take
> responsibility for doing this a bit more explicitly (or calling
> normalizeDocument when it's done -- which is probably a less efficient
> option),
yes we should make it clear when these defaulting and whitespace normalizing
do occur.
should we consider having an option to turn off these defaults and
normalizations completely even when validating?
my reasoning is that we probably will NOT want to use validation on every XML
message we receive (because of the high CPU cost) so when a system is running
in a deployed environment we will probably turn off validation to get an extra
30% throughput. The problem is that turning off validation can change the
value of the DOM tree and (default values, whitespace normalization) and so
intern change the behavior of the program that uses the DOM tree.
Could we standardize a mode where we validate but don't ever add values or
change the whitespace (so the program won't be affected by the validation
setting)?
> and (b) the DOM would not assist you in maintaining those defaults
> as you edit the document -- if you remove a Specified value the default
> would not magically re-appear until/unless you re-normalized.
>
> I belive this change is being proposed partly because (b) is much harder to
> support in the Schema world than it was with DTDs. Giving up the magic
> generation of default Attrs does seem to be the simplest fix.
Yes it does, I am fine with the change.
Cheers
David
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
