Lists Home |
Date Index |
> > But based on this I think that NVDL, without an api to request the
> > validated stream, will not be seen as useful by the wider community.
> Well, then whole Java world has a really big problem, because Java
> validation API introduced in Java 1.5 (JAXP 1.3) performs only
> validation, nothing more is returned. JNVDL simply uses this API,
> because it is standard way how to invoke validation.
I'm sure it is, it's the standard way for most validation languages,
however these languages tend to assume that the complete validation is
done in the language. NVDL does not do the validation in the language,
it passes it to other validation engines, basically.
Let's suppose I'm out talking to some people, they tell me: hey, we
want to extend our xml format a with some stuff from xml format b, but
to do that we need to edit the schemas and that's a lot of work.
Then I say: oh no, you can just write an NVDL script, it will pass the
xml in format a to your schemas for format a and your xml in format b
to your schemas for format b.
This sounds great, only I know that what they actually want is not
just that validation, but validation, seperation of the namespaces
that have been tangled, and sending of these things to different
So if I tell them, well you can seperate your namespaces out very
easily this way, and send them to different processes, then the
natural response of the guy in charge on their side is, okay, once
I've seperated stuff out why don't I just send namespace a to schemas
for a and namespace b to schemas for b? What value do I get from NVDL.
And it is at that point that I'm not sure what I should tell him.
> Do you have any idea how API that returns outcome of NVDL validation as
> validated streams your are talking about should look like? I still don't
> get it.
It does not actually have to be streams, I just got into talking
streams cause it seemed someone else was thinking specifically of
streams, perhaps this made things more difficult to communicate. But I
would suppose an api would expose something like the following:
Nvdl.Instance.Namespaces - an array of namespaces
Nvdl.Instance.isValid(namespace) - boolean
no doubt there would be other things but with this then one could do,
as per following pseudocode
for each namespace in Nvdl.Instance.Namespaces
Even something as trivial as this would, I believe, go a long ways to
increasing the sellability of NVDL for the hypothetical project
leader/senior developer type I described above.