[OASIS Issue Tracker] (OSLCCORE-9) Bootstrapping discovery

[OASIS Issues Tracker <workgroup_mailer@lists.oasis-open.org>]

    [ https://issues.oasis-open.org/browse/OSLCCORE-9?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=60671#comment-60671 ] 

Jim Ruehlin commented on OSLCCORE-9:
------------------------------------

Is the use case we're looking at here as simple as "Do a GET to a unique URI (possibly with a custom header) and get a URI to the Service Provider Catalog"? Any way you look at it, we must start with a unique URI. The idea scenario would only require that much information to get everything else you need from an OSLC server.

I'm wondering if adding requirements for a new vocabulary like VoiD, and/or making the implementer parse the results of that new vocabulary, is too much overhead just to grab the initial URI. It seems like less of a burden to have the implementer just find the URI they need.

I haven't heard complaints about finding a base URI for an OSLC server. But there are general complaints about the complexity of OSLC, the vocabularies it depends on, and the inconsistent way implementers initially connect to OSLC services (e.g. root contexts and other non-OSLC constructs). So it seems like keeping the initial OSLC handshake very simple and consistent would be a good improvement.

If my reasoning is correct, then it seems like an OSLC server MUST provide a unique URI that can be queried to establish domain support and to get the URI to the service provider catalog. Then if you have that initial URI you've got the keys to the kingdom.

> Bootstrapping discovery
> -----------------------
>
>                 Key: OSLCCORE-9
>                 URL: https://issues.oasis-open.org/browse/OSLCCORE-9
>             Project: OASIS OSLC Lifecycle Integration Core (OSLC Core) TC
>          Issue Type: Task
>            Reporter: Martin Pain
>            Assignee: James Amsden
>              Labels: ready-for-vote
>
> See https://lists.oasis-open.org/archives/oslc-core/201503/msg00016.html
> The problem is about the user experience's when configuring OSLC clients to connect to OSLC servers. I think having to search a server's documentation and UI to find a discovery URI that is used only once is not a good user experience, and this will come very early in new users' experimentation with OSLC (both developers and administrators).
> The resource the client wants is the "root" ServiceProviderCatalog. The user will be aware of the idea of URIs representing resources, but the resource in their mind is the server, and the URI they associate with the server is most likely the home page of its web UI. So I suggest we define a Link header rel that is used to link that web UI URI to the ServerProviderCatalog URI.
> So then the user's experience will be:
> - They are reading the client's instructions for integrating with OSLC servers.
> - It says to enter the URI to the server's home page.
> - The user copies that from their browser and enters it into the client.
> - The client finds what it needs on the server. (Under the covers: they do a OPTIONS request on the URI that was provided, follow the Link header that has the rel that we define, and request that URI with an Accept header for the content types that they support, and receive the root ServiceProviderCatalog).
> (The original proposal has been moved to a comment on 24th July 2015 to avoid confusion with the revised proposal that is in the "proposal" field above.)



--
This message was sent by Atlassian JIRA
(v6.2.2#6258)