James M Snell wrote:

Aleksander Slominski wrote:


it is interesting that atom reader (in this case it seems to SharpReader) is actually rendering text content of

|<content type="application/xml">
<EndpointReference xmlns="http://www.w3.org/2005/08/addressing";>
     <Address>http://localhost:9085/wsatom/services/HelloWorld
</Address>
     </EndpointReference>
   </content>|

are atom readers supposed to do that? would it not be better to have a proper XHTML as content and store EndpointReference directly as a child inside atom:entry? after all EndpointReference is used by services and not Atom GUI apps?

If you take another look at the feed generated, I actually put text content in the summary element. Unfortunately, SharpReader seems to ignore it in favor of the content. I think that as a best practice, Atom readers should defer to the summary if the content type is anything but Text, HTML or XHTML.

good point. however that also depends on what is the intended use for the feed: if it is only used for WS introspection then it is targeted for machine processing and maybe administrators that need to see it but not ordinary users/developers but there may be some other usage patterns where a feed will contain entries mixed with regular blog entries. for example xmethods.net could publish not only all service endpoints (WSDL or WSA EPRs - and that would be one long list unless feed history is used) but it may as well publish news feed with new services (and their associated WSDLs, EPRs, ...) along with general changes to the website, hosting, editorials, etc. such feed would be targeted for developers but also easy to use for WS discovery.


second comment is very difficult (or impossible) to know that feed is actually representing "WS introspection" - should there not be there a some (optional) element inside atom:feed to indicate that feed contains (only) list of EndpointReferences?

I've been considering a profile element that can sit on the feed and entry level to indicate this sort of thing. Essentially, the profile element would be used as a key to indicate what kind of feed/entry this is.
For example,
 <feed xmlns="http://www.w3.org/2005/Atom";>
    <x:profile>http://example.com/atom/profiles/ws-inspection</x:profile>
    ...
 </feed>

One could easily imagine profile identifiers for blogs, podcasts, news, lists (ala netflix queues), etc.

maybe that should be a special kind of atom:link with rel="http://example.rog/Atom/feedType|feedProfile"?

maybe it would be good to allow multiple feed types to apply to one feed (or maybe not as it starts to sound like multiple inheritance in OO ...)

best,

alek

--
---         Memorable Quote from Firefly (105. Out Of Gas)
-Ship like this, be with you until you die -That's because it's a deathtrap

Reply via email to