Todo List

Member ast::ParallelMap::ParallelMap (AstCmpMap *rawptr)

add a test that the CmpMap is parallel

Member lsst::afw::detection::FootprintSet::FootprintSet (FootprintSet const &footprints1, FootprintSet const &footprints2, bool const includePeaks)

Implement this. There's RHL Pan-STARRS code to do it, but it isn't yet converted to LSST C++

Class lsst::afw::image::pixel::BinaryExpr< ExprT1, double, ImageBinOp, MaskBinOp, VarianceBinOp >

Could use a traits class to handle all scalar types

Class lsst::afw::math::BilinearWarpingKernel

: make a new class WarpingKernel and make this a subclass.

Class lsst::afw::math::LanczosWarpingKernel

: make a new class WarpingKernel and make this a subclass.

Class lsst::afw::math::NearestWarpingKernel

: make a new class WarpingKernel and make this a subclass.

Member lsst::afw::math::SpatialCell::visitAllCandidates (CandidateVisitor *visitor, bool const ignoreExceptions=false, bool const reset=true) const

This is currently implemented via a const_cast (arghhh). The problem is that SpatialCell::begin() const isn't yet implemented

Member lsst::afw::math::SpatialCell::visitCandidates (CandidateVisitor *visitor, int const nMaxPerCell=-1, bool const ignoreExceptions=false, bool const reset=true) const

This is currently implemented via a const_cast (arghhh). The problem is that SpatialCell::begin() const isn't yet implemented

Member lsst::afw::math::Statistics::getResult (Property const prop=NOTHING) const

uncertainties on MEANCLIP,STDEVCLIP are sketchy. _n != _nClip

Member lsst::afw::math::warpImage (DestImageT &destImage, geom::SkyWcs const &destWcs, SrcImageT const &srcImage, geom::SkyWcs const &srcWcs, WarpingControl const &control, typename DestImageT::SinglePixel padValue=lsst::afw::math::edgePixel< DestImageT >(typename lsst::afw::image::detail::image_traits< DestImageT >::image_category()))

Should support an additional color-based position correction in the remapping (differential chromatic refraction). This can be done either object-by-object or pixel-by-pixel.

Need to deal with oversampling and/or weight maps. If done we can use faster kernels than sinc.