------------------------------------- Summary of BDF Review Telecon 7apr08 ------------------------------------- M. Rupen 7apr08 (checked by B. Glendenning) On-line: Bryan Butler, Lindsey Davis, George Moellenbrock, Steve Myers, Jim Pisano, Martin Pokorny, Michael Rupen, Ken Sowinski, Francois Viallefond, Manabu Watanabe This is mostly a list of DECISIONs and ACTION ITEMs. Note Section V giving the next set of dates and "high-level" action items. I. What version is documented? DECISION: This version of the BDF definition covers the ALMA-B correlator and will be used at the ATF. It is NOT compatible with the ACA. - The main difference is that Zero Lags are *optional* for the ACA; various minor changes follow from this. ACTION ITEM: add a note stating that the ACA correlator is not fully included in this document, but will be covered by the next revision, due 15may08. [JP] DECISION: The ASDM Finalization FBT will indeed use schema version 097 (with revisions in line with this BDF document). Note that the BDF document under discussion here is more nearly in line with 096; see below for details on producing the schema aligned with the BDF defined here. II. Highlights and unresolved issues from the BDF team's response 1) The description of the size of the "POL" axis will be extensively revised. [MP] -- OK 2) The description of the "dimensionality" and "numTimes" elements will be extensively revised. [MP] -- OK 3) Zero lag components, elements and attributes will be made fully optional to support the ACA correlator. -- DECISION: postponed to the next version 4) Rationalization of data types will be deferred to a later BDF version. -- DECISION: postponed 5) A glossary will be provided. -- OK [JP] [MP] 6) Baseband names, spectral window names, sideband and sideband image ids: should they occur in the BDF? -- DECISION: these will be left "as is" for now -- ACTION ITEM: include a fuller description of the sideband attributes etc., making it clear that the BDF elements are describing what is done in separating the sidebands for a double sideband (DSB) mixer in the correlator/backend, rather than keeping track of the sideband observed with a single sideband (SSB) mixer [JP] 7) Is an "UNKNOWN" value required for "CorrelatorName" enumeration? -- DECISION: No. Keep the enumerated list as-is; add correlators to the enumeration as needed. -- Some discussion of whether this might better be a string. -- ACTION ITEM: Mention that this value is not used currently by ALMA. [JP] 8) Sideband (pair) identification: should it be in the BDF? -- DECISION: Leave it in. -- ACTION ITEM: add a fuller description -- see above [JP] 9) Are elements of ZERO_LAGS always real-valued, always complex-valued, or can the choice be BDF-instance dependent? -- DECISION: ZERO_LAGS are always real, since we only care about the values for the parallel hands. -- ACTION ITEM: add a note to this effect, with explanation. [JP] 10) Should baseline ordering (BAL axis) be changed? -- DECISION: No, leave this as-is for now. The ordering makes as much or as little sense as the "normal" order of visibilities; the filler will move things around if necessary. The key is to agree on an ordering of visibilities, and describe it unambiguously. -- Various people (FV, KS, etc.) noted that the present ordering is more historical than logical. However, since some software already produces/reads this ordering, and there were no strong arguments for changing it, the DECISION was to stick with what we've got. -- FV suggested adding a parameter stating the order explicitly, since we basically have two choices (the usual way and the current-BDF way). DECISION: This again seemed too much like (unnecessary) work. 11) Should ACTUAL_TIMES be stored as a time difference relative to time stored in the SDM to save space? -- DECISION: No. Adds extra work for little or no gain. 12) Should execBlock element be optional (rather than mandatory)? -- DECISION: Yes, unless this is required for the filler (see Action Item). -- ACTION ITEM for BG: Ask Michel whether the filler needs this element. If YES, leave it as mandatory; if no, switch to optional. [MP] -- ACTION ITEM: add note giving the reasoning behind putting this in the BDF (for access to data without going through SDM, e.g., to look at the data with various real time displays). [MP] III. Additional issues from the Long List 1) Can auto-corr'ns be stored with cross-corr'ns or not? [Item 9] [MP] -- DECISION: No. ALMA stores them separately to allow different data types (integer vs. float); EVLA will not take advantage of that flexibility, sticking with float for both, but there is no strong argument for *not* storing the auto-corr'ns separately. Note that this decision refers to the actual data, not the metadata. 2) ANT+BAL size [item 11] [MP] -- Auto- can't be store with cross-corr'ns (see above) in the actual data, so ANT+BAL is not allowed for the actual data. -- ANT+BAL *is* allowed for the metadata (times, flags, durations). 3) Holography [item 16] -- DECISION: Holography is both optional and not fully defined [per BG]. -- ACTION ITEM: Add an ALMA note describing what "holography" means in this context (using a reference horn on a single antenna, i.e., single dish holography). [JP] 4) NetSideband [item 21] -- covered above [JP] 5) AUTO_DATA real vs. complex [item 34] -- covered above [JP] 6) Meaning of "non-standard" [item 62] [JP] -- DECISION: Eliminate this term. "Non-standard" here actually means "ALMA will never produce CROSS_ONLY data". -- ACTION ITEM: change the wording to reflect this. 7) Size of weights [item 109] [MP] -- ACTION ITEM: drop TYPE attribute for weights -- ACTION ITEM: include a formula/description for computing the total size of the weights table