On Mon, 2010-04-12 at 10:28 +0200, Michal Fojtik wrote:
On 09/04/10 16:59 -0700, lutter@redhat.com wrote:
XML Representation
The XML representation for hardware profiles looks like
<hardware-profile href="http://localhost:3001/api/hardware_profiles/m1-large">
Why we use both hardware-profile and hardware_profile ? This must be handled by extra code in client/aggregator. For me, using hardware_profile also in URL is fine (same for storage(-_)snapshots/volumes).
An attempt at foolish consistency ;) I thought that we use '-' to separate words in most tag names, rather than '_'. But that's not true, and we do use '_' e.g. in <owner_id/>
Based of the discussions with the RHEV-M guys, it might be best if we take another look at all our XML, and replace '-' with '_' for consistency.
I think though we should just do that with another cleanup patch.
Nice. We should consider creating a DTD or XScheme for all XMLs we produce and publish it on website (+ integrate it as part of tests)
Agreed; I am a little hesitant to do that, since writing good schemas is a lot of work, and only worth it if anybody pays attention to them (and in any event, it should be Relax-NG, DTD aren't all that expressive, and XML Schema is inscrutable)
Nice. Why isn't 'id' part of element btw? According to this whitepaper (http://www.ibm.com/developerworks/xml/library/x-eleatt.html) 'ID' is perfect candidate for element attribute.
Another thing that I'd like to see cleaned up, though there's some discussion around using a consistent style over all REST API's we publish (e.g., for RHEV-M, too) But you are right: <id/> should really be an attribute.
David