Re: [gobo-eiffel-develop] Merging clusters in XACE

[Eric B. <er...@go...>]

Alexander Kogtenkov wrote:
> Eric Bezault wrote:
> 
> 
>>      <mount location=".../library.xace">
>>          <cluster name="a">
>>              <cluster name="b">
>>                  <cluster name="c">
>>                       <option name="assertion" value="require"/>
>>                  </cluster>
>>              </cluster>
>>          </cluster>
>>      </mount>
>>
>>I think that after writing it a few times I start to get
>>used to this notation despite the cluster hierarchy repetition.
>>Especially if we get an error if this hierarchy does not
>>match the one found in the mounted library.
> 
> 
> Is it OK to add a cluster into an existing cluster hierarchy?

I would say no. I'd rather prefer having an error because
the cluster does not exist (anymore) in the mounted library.

> E.g., we have a cluster structure like above and want to
> add a new cluster "d" inside "a". Such a possibility
> can be used to distribute libraries in parts. For example,
> Gobo XML is useful even without XPath. But when required,
> a new XACE file "adds" XPath cluster inside XML cluster.
> If this is allowed, how to specify it?

But would we add Xpath cluster(s) or mount a library.xace
describing the Xpath cluster(s) once and for all? Do the
Xpath clusters need to be subclusters of XML clusters when
we want to use them? Can we have a library mounting the
basic XML library and the XPath extension library if we
want to mount everything at once?

Another solution would be to have a library.xace with
everything in it, and being able to ignore some clusters
(and their subclusters) when mounting it.

-- 
Eric Bezault
mailto:er...@go...
http://www.gobosoft.com