Re: writing collection of refobjects through xmiwriter

Hi Martin,

> Hi Brian,
>> need the getLocation() method. But, if you want more flexibility (e.g. 
>> the ability to do full XPointers or some other format), then you would 
>> need the getLocation() method to do that.
> 
> do you think that this is important? i would rather have only two 
> methods - getDocumentURI and getXMIID and write hrefs in form 
> <documentURI>#<xmi.id>. I think this format is a standard format for 
> hrefs in XMI. Do you think it is too limiting?

No, not for me. I believe that people on the XMI RTF said that requiring 
support for any other format of link was asking too much when they were 
discussing XLink this year or last year. So, I think this format (called 
the "bare names" format in the XPointer spec) is all that is needed. 
(Note that some of the XMI diff files on the MDR file download site are 
using links in the format <document-uri>|<xmi-id>, which was an older 
syntax proposed for XLink a long time ago).

Also, in the message I just posted in response to Ronald, the I 
suggested a link in a format #xmiid[type-name] which isn't compliant 
with any standards AFAICT. But to get some kind of standards-compliance 
for what Ronald wants to do, he would have to encode the type-name in 
the document URI. That means he would no longer be able to determine if 
objects where in the same XMI document by comparing the document URI's. 
But, personally, I don't like the idea of encoding real information in 
the URI's in the first place, because that hurts interoperability (no 
other tools would understand the URI's) and XMI is supposed to be an 
interchange format.

- Brian