[cddlm] CDL xsd

[Guilherme Mauro Germoglio Barbosa / Projeto Ourgrid - Projeto Ourgrid/LSD]

Well, after testing some other cases, I noticed that the patch leaves 
some other problems considered as valid. =(
But the idea is to specify that the only root element allowed is 
<cdl:cdl> - something that decreases the reuse of the schema, as some 
authors say. I will try to find a better way to make this work.

I've also noticed that /test/cdl/valid/cmp/cmp-switch.cdl and 
cmp-flow.cdl are not well-formed.

Guilherme


Steve Loughran wrote:

> Guilherme Mauro Germoglio Barbosa / Projeto Ourgrid - Projeto 
> Ourgrid/LSD wrote:
>
>> Hello all,
>>
>> I've been looking at your repository and I found out that some cdl 
>> files located in /deployment/test/cdl/invalid/ are considered valid 
>> according to the cdl schema in the repository.
>>
>> One of them is /deployment/test/cdl/invalid/wrong_root_elt_type.cdl 
>> /that has its root element /<cdl:expression> /instead of <cdl:cdl> as 
>> it must be.
>> I think the CDL XMLSchema (xml-cdl.xsd) should care about 
>> invalidating this kind of error - I modified the current schema to 
>> invalidate this.
>
>
> would you care to submit a patch?

>
>> The other is 
>> /deployment/test/cdl/invalid/expression-duplicate-variables.cdl /- 
>> unfortunately this one we still don't know how to invalidate through 
>> the schema.
>>
>> So my question is if the invalid files from the repository must be 
>> invalidated simply by XMLSchema or the parser implementation should 
>> care about these errors.
>
>
> Yes, some of the stuff in invalid/ is invalid semantically, not 
> syntactically. There are limits to XSD's powers. As I write more of 
> the import and resolution tests, we'll have docs in there that are 
> invalid there to, for example
>
> Import tests
> -multiple imports of the same doc into different namespaces (is that 
> invalid?)
> -multiple imports of different docs into the same namespace
> -docs that dont resolve to valid urls, even on the classpath
>
> resolution
> -things that dont resolve
>
> extends
> -self referential extension
>
> others
> -conflicting attributes
> -expressions that dont make sense
>
> etc. etc. I'm just adding tests as i go along. We could structure them 
> better, i suppose. I am glad that you have embraced testing!
>
>

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3186 bytes
Desc: S/MIME Cryptographic Signature
Url : http://www.ogf.org/pipermail/cddlm-wg/attachments/20050610/1ed14e52/attachment.bin