From: Configuration Control Board

CCB Meeting 1 October 2008

Attendees

Robyn Allsman
Robert Lupton
Ray Plante
Tim Axelrod
Russell Owen
Kian-Tat Lim
Jeff Kantor, briefly

Proposals Discussed

Meeting

Open Source License

Ray Plante provided last month's CCB's decision to Jeff.

Robert felt that his web reading on our concept of having a 2 tier license (source is Apache-2; binary is GPL-3 if includes any GPL-3 binary) is not legally possible. His understanding is that if the source requires a GPL-2 library, then the source is also GPL-2. Also, even though Apache-2 is compatible with GPL-3, if the source requires any GPL-3 library, then the source becomes GPL-3 and cannot remain Apache-2. K-T and Russell objected that their understanding is quite different: that the LSST source remains under whatever license is chosen, regardless of the use of other packages, while the binaries will be GPL if any GPL-3 libraries are used; GPL-2 libraries cannot be used with Apache-2 source. Robert emphasized the need for a legal review (although there was some concern that an AZ legal review was marginally adequate).

Jeff is presenting to the LSST Board on Friday a proposal for DM Open Sourcing with a recommendation that it be submitted for legal review.

Robyn mentioned that the permanent URL for the hosting of Copyright and other legal notices has been setup (http://www.lsstcorp.org/LegalNotices/).

Pixel Coordinate System

Robert said the proposal had a narrow scope to define the absolute reference point for an image. This was brought out when Tim asked how multiple CCD segments could be addressed using this proposal.

Russell suggested that, if the bands between segments were integer pixel size, then a mask could be created which reflected the bands so that original proposal would remain accurate.

Tim has accepted an action item to develop a Proposal on how coordinate systems of individual segments within a CCD can combine into a single global coordinate system. This is recorded in the Proposals for Future Review as: "defining a single global coordinate system across individual CCD segments".

Russell recommended the mandatory use of functions to translate between pixel addressing schemes such as index -> position, position -> index, fits -> index, etc. He has accepted an action item to develop a Proposal on "Standard Use of Pixel Addressing Conversion Functions". Note that Robert prefers to optimize his code by direct integer addition instead of stepping through a function when he knows that a simple integer translation offset will suffice; Tim prefers consistent use of address translator functions.

Boost Library inclusion

Robert said he'd reviewed the use of Boost libraries in the DM code but hadn't updated the Boost web page. [ Ed note: shortly after the meeting he updated it. ]

He found one TR1 hash table (unordered map) provided by G++ but not in the Boost TR1 compatibility library which would need a wrapper (which has already been written).

Status of Past Action Items

See: CCB Action Items for the complete list.

  • Ray reviewed the Build system status. He will merge his changes onto the trunk after the CCB meeting. He will document the architecture of the build system; will summarize the changes required for distributing packages from the central site's perspective; and will write up how developers initiate creation of new distribution packages. Further technical discussion was directed to another MW/DC3 meeting.

New Action Items

  • Russell Owen will prepare a Proposal on "Standard Use of Pixel Addressing Conversion Functions".
  • Tim Axelrod will prepare a Proposal on "Defining a single global coordinate system across individual CCD segments"
  • Robert Lupton and Ray Plante will writeup their combined document on the Build System. Ray providing documenation on the Build system's architecture and Robert documenting how developers will most effectively use the system.