Time:
1:00 PM Eastern US (contact
MichaelFiedler if you'd like to participate)
The Automation meetings alternate times each meeting to accomodate the global team.
Agenda
* Reoccurring agenda items:
* Main agenda items:
-
- Report to workgroup on artifact discussion (Automation Request/Result vs single AutomationJob? )
- DavidBrauneis, CharlesRankin, LucasPanjer and MichaelFiedler met to discuss
- Summary
- Some existing implementations represent the concepts with 1 artifact, some with 2
- 2 artifacts provides the most flexibility
- implementations with only 1 artifact internally can represent the information in that artifact as 2 separate OSLC artifacts, the request and result.
- keep the request as simple as possible. Minimal info to request an automation:
- state
- desired state (for cancel)
- input parameters required for automation execution
- link to result (when it becomes available)
- link to automation plan
- requester
- time requested
- possibly "how requested" (manual, scheduled, programmatically)
- originally, discussed this being transient. Now recommended persistent. Implementation may expose info available in 1..n places as 2 OSLC artifacts.
- Discuss Agent/Worker sub-scenario.
- Proposal to limit scenario to optional agent registration
- Value of a limited scenario
- If there is time, discuss result contributions. See ResourceDefinitions.
- Plans for moving from scenario development to spec development
- Previous Action Items:
- Next meetings:
- Discuss meeting time
- 1 December
Minutes
Attending: Michael Fiedler, Rich Rakich, Charles Rankin, Vaibhav Srivastava, Pramod Chandoria, Paul
McMahan? , Bill Higgins, Petes Steinfeld, Barys Dubauski, Max Vohlken
* Discussion topic 1 was the review of Automation Request and Automation Result as two separate artifacts (vs Automation Job)
- General consensus
- Going forward, we will use these two artifacts in addition to Automation Plan and the as-yet undiscussed contributions
* Second topic splintered off from the first. Centered around whether the notions of composition and step ordering were in scope for V1
- Agreement that the scenarios around composite execution are very important
- Some disagreement around how much is in scope for V1
- Agreement it is still an open issues
- Many examples from build domain to illustrate importance of ordering. Simple result contributions may not suffice
- MaxVohlken feels we may be missing a result container artifact
* Actor registration discussion
- Similar to the second topic above, much of the discussion resolved around scope rather than the specifics of this scenario
- Acknowledgement that the V1 spec needs to fit the in-scope V1 scenarios
- Discussion around need to get initial implementations to find what is missing, what is too much
- PaulMcMahan introduced actor registration as a means for setting an automation system in motion
* Need to move into spec authoring mode
- We still have scenarios to discuss (result contributions, DevOps? etc)
- In parallel, need to start spec authoring. CharlesRankin has volunteered to help. Others interested, contact MichaelFiedler.
- Need to focus on resources and properties...what is inherited from OslcCore and what is unique to Automation.
* Next meeting: Thursday, 1 December at 10 AM Eastern US time