Meeting began   @ 10am; ended 11:50pm Project Time.

Attending

Next meeting:  @ 11am.

New Use Cases

Tim Jenness reported that he had added two new Use Cases (ARCH3 and DAX6) relating to raw data retrieval and augmented headers. This triggered some discussion asking for clarification. Three types of "raw" data were described that could be useful to different Actors:

  • raw data as written (expected to be in FITS format) on the night.
  • raw data with any incorrect headers fixed up and any headers deemed useful after the fact now included. There are cases, for example, where later in the survey we determine that some extra piece of information from the EFD would help reprocessing, or where we realize that a particular sensor has a calibration error.
  • raw data augmented by information determined by earlier reprocessing (either in the L1 system or as part of a data release). For example, accurate WCS information.

It was noted that if "raw++", as we were calling it in the meeting, is to be supported then it would be important to be able to retrieve a version of the raw file with the augmented headers as they were for a particular data release.

EFD queries were discussed. Everyone is happy with the concept of "pre-flight" being able to specify the EFD query that should be used to access the subset. Rather than specifying specific channels of data in a configuration file, Jim Bosch felt that it might be possible to have named EFD subsets that cover groups of related channels which would then be subsetted in time by the exposure start/end information.

Brian Van Klaveren reported that he had discussed Use Cases with the DAX team and did not feel that any more should be added from them.

Russell Owen reported that he had completed the L1 Use Cases and included some extra ones he felt were important.

Requirements

Now that the Use Cases are in a relatively good shape using standard Actors, we moved our focus to requirements gathering.

Unknown User (pschella) wondered if we can use "The Butler shall" in a requirement. After some discussion we decided to agree on a set of definitions from DMTN-056 with the addition of three distinct concept for "Butler":

  • I/O System (or Input System and Output System if only one way) covering data persistence.
  • Transfer System covering data replication and subsetting.
  • Discovery System covering the discovery of datasets based on partial data specifications or relating to other datasets.

Everyone was tasked with writing up 5 requirements based on their Use Cases before the next meeting.

  • No labels