Details
Description
LDM-592 contains the use cases developed by the Butler Working Group in August. This document has to be reviewed and approved since it is a foundational document for the Butler requirements. I've allocated 2 weeks for this.
The document can be found at: https://docushare.lsst.org/docushare/dsweb/Get/Version-48456/LDM-592.pdf
There is a pull request on github where you can make comments on the text: https://github.com/lsst/LDM-592/pull/1
Attachments
Issue Links
Activity
I think we should publish and ticket changes. Lets give Michelle Butler and Margaret Gelman until Monday to see if they want to comment on the outstanding use cases.
Here are the combo comments from Don, Margaret and I. Thanks Wil for giving Margaret and I another day to get comments in.
A few high-level concerns from the Butler use case and requirements review, from the operations and production computing viewpoint:
- Prioritization of requirements is needed. Which use cases are really important and MUST be supported day 1 for operations, and which ones would be nice, and will be added in time…or have a phased approach.
- Butler is planned to do all things like managing provenance, data location, and data transfer, and hiding file formats, and to do so for all use cases in the project (development, test, integration, production, end user analysis, etc.). This is a substantial undertaking, especially for bespoken code.
Our understanding is that additional use cases are expected to be added to this document after initial baselining, based on the findings during the Butler design prototyping. So, baselining the document as-is is likely adequate for now.
Michelle Butler
added a comment - Here are the combo comments from Don, Margaret and I. Thanks Wil for giving Margaret and I another day to get comments in.
A few high-level concerns from the Butler use case and requirements review, from the operations and production computing viewpoint:
Prioritization of requirements is needed. Which use cases are really important and MUST be supported day 1 for operations, and which ones would be nice, and will be added in time…or have a phased approach.
Butler is planned to do all things like managing provenance, data location, and data transfer, and hiding file formats, and to do so for all use cases in the project (development, test, integration, production, end user analysis, etc.). This is a substantial undertaking, especially for bespoken code.
Our understanding is that additional use cases are expected to be added to this document after initial baselining, based on the findings during the Butler design prototyping. So, baselining the document as-is is likely adequate for now.
Thanks for the comments. Requirements are prioritized in the requirements document (LDM-556) and we haven't really tried to prioritize Use Cases before. I'll release the document today.
Now on docushare at https://docushare.lsst.org/docushare/dsweb/Get/Version-49605/LDM-592.pdf
I have implemented the requested use cases from Simon Krughoff, Gregory Dubois-Felsmann, and Robert Lupton.
I am still awaiting feedback from NCSA, and use cases from Brian Van Klaveren for round-tripping a repository upgrade, and from Gregory Dubois-Felsmann for preserving old data and code.
Wil O'Mullane (or any one else from CCB) should I wait for these use cases and comments (they were requested as part of the review) or should I baseline what we have and add the new use cases later (converting them to tickets for Brian and Gregory)?