• Add a Comment
  • Edit
  • More Actions v
  • Quarantine this Entry

Comments (1)

1 localhost commented Trackback

Hi Simon,

 
thanks for your thoughtful post and thanks for starting
the conversation.
 
I think I completely appreciate the usecase of PUTting content
into an empty repository to quote your example "/a/b/c/mydoc.txt"
and all the issues around not having to come up with types
before you put something into the repository.
 
I am very much a "Data First" person as you may find here:
http://wiki.apache.org/jackrabbit/DavidsModel
 
While WebDAV for example does not allow to PUT or MKCOL
resources that are not direct "children" of existing collections of
course HTTP does and I fully agree that this very convenient
for the user. This is also why we implemented it Apache Sling
(our web application framework based on REST principles, currently
in the incubator) the creation of arbitrary new resources in
exactly the fashion you describe. Sling is entirely JCR backed.
 
Even in JCR on the repository level the application needs to
deal with situations where a node exists without a readable/visible
parent and the fact that our Node.addNode() methods take
a "path" rather than a "name" (path segment) should indicate
that an implementation potentially could allow for a
non-contiguous creation of content.
 
I think from a JCR specification process we would certainly
be very interested in active participation in case you find
that JCR prohibits something that you would like to pursue
in an implementation, and I think historically the JCR expert
group has been very open to accommodate features like that.
 
I think the amount of shared goals between JCR/Jackrabbit/Sling
and JRS are astonishing and I would be very interested in being able
to show you the interesting overlaps. Feel free to shoot me an email
in case that is of interest for you.

Add a Comment Add a Comment