From: Configuration Control Board

Table of Contents

  1. CCB Meeting 5 November 2008
    1. Attendees
    2. Proposals Discussed
    3. Meeting
      1. Defining a single global coordinate system across individual CCD Segments
      2. Standard Use of Pixel Addressing Conversion Functions
      3. Boost Library inclusion
      4. Unit Testing Standard
      5. Guideline for Adding or Moving SVN Directories
      6. So You Want to Add a New DM Baseline Requirement…
    4. Status of Past Action Items
    5. New Action Items

CCB Meeting 5 November 2008

Attendees

Robyn Allsman
Robert Lupton
Tim Axelrod
Russell Owen
Kian-Tat Lim
Roc Cutri

Ray Plante was attending ADASS.

Proposals Discussed

Meeting

Defining a single global coordinate system across individual CCD Segments

Tim discussed his concept for a global coordinate system accross CCD segments. He noted it is only useful within a single piece of silicon and is not extendible to the entire focal plane.

He noted that Russell's pixel coordinate conversion functions would become class members in his concept. Russell agreed.

KT was concerned about potentially large discontinuities in a ccd. Tim didn't think this was a show-stopper issue; the 'discontinous bit mask' should handle the breaks. Robert reviewed the major algorithms (ISR and Image Detection) and determined they would not be broken by the scheme but might need special handling in specific well-known situations.

KT wondered if the concept could be extended to rafts? Tim hadn't considered this but does agree this would be useful when visualizing the entire raft. A discussion on the current 1:1 mapping of the ccd to the maskedimage vs the potential 1:n mapping of an alternate scenario was discussed.

Russell wondered how the overscan region would be handled. Tim said the raw image contains the overscan in the pixel grid; ISR creates a new image with the overscan trimmed.

Tim noted that work remains to make the concept complete. Robert pointed out that this 'proposal' was really an implementation issue and not really a TCT (aka CCB) issue and should be handled in the Applications WG. We all agreed to push this proposal to that WG.

Standard Use of Pixel Addressing Conversion Functions

Russell reviewed his proposal to mandate use of pixel coordinate conversion functions when translating between coordinate systems.

Robert feels that excessive use of the conversion functions could obfuscate the code.

Roc was concerned that including this requirement would change the current task dependencies and lead to DC3 milestone slippage since there would a lead time between the implementation of Tim's global coordinate system and its specification of the conversion functions. Robert felt the existing conversion functions could be used in the interim and when the new class based conversion functions were ready, the compiler would flag all code needing revision and the transition would not impact the schedule.

All TCT members present agreed to the requirement to use the conversion functions.

Boost Library inclusion

Robert is mapping our current use of Boost libraries to both TR1 and TR2. He will update the boost? webpage soon.

Unit Testing Standard

Robyn reviewed the unit testing procedure and its use of coverage analysis tools.

Robert and Russell were concerned that the current unit testing scenario used by the developers is a hybrid where python:unittest wraps the C++ objects under test. Robyn will revist the procedure to determine if such a hybrid use of gcc coverage instrumentation works.

Russell pointed out the example of LSST versioning of vendor products included blanks which are not allowed in the LSST versioning specification.

Proposal was sent back for revision.

Guideline for Adding or Moving SVN Directories

Robyn discussed the layout of the SVN repository and reasons for its current structure.

Robert wanted assurances that the current structure is enduring since the DC2 to DC3 reorganization was tedious. Since the new organization is based on the WBS layout which is designed for the Construction phase and beyond, there is no expectation of a future major revision.

Proposal was passed.

So You Want to Add a New DM Baseline Requirement…

Robyn took an email from Jeff Kantor discussing the procedure to follow should a new Baseline Requirement be proposed and put it into generic guideline form.

Roc suggested additions to clarify the process should the DM Manager prohibit the baseline addition due to funding reasons (i.e. present new Science requirement to the LSST Science Council who are empowered to revise the LSST Science Requirements which would flow down to the DM Functional Requirements).

Proposal was sent back for revision.

Status of Past Action Items

See: TCT Action Items for the complete list.

  • 64 bit build available.
  • Contract for SWIG developers to develop 'Standard definition augmented assignment operators' is no longer needed since the issue was resolved using the *.i definitions.
  • Proposal for unit test's coverage analysis tools submitted but needs refinement.

New Action Items

  • Robert Lupton - will ask Jeff Kantor about possibility of setting up a retainer contract for SWIG development for LSST needs.