= Scantable vs MS =

== Issues ==

 * changing ASAP to use MS underneath means rewriting almost all methods
   * everything is based on row/Vector operations
 * Scantable stores polarisations individually, MS stores them in a matrix
   * no referencing possible - requires copying
 * Scantables stores Coordinates (per band), MS stores per channel information
 * MS is overkill for SD data

== Solutions ==

 * change Scantable schema closer to MS to allow future move.
 * write direct MS -> Scantable filler
   * only copy when necessary
   * faster
   * both  are based on casa::Table so clever things can be done
   * all non-asap relevant information could be copied/referenced
 * write Scantable -> MS writer
   * preserve all (most) information/meta-data
 * don't use MS, use ALMA to Scantable Filler ;-)
 * trial implementation of complex functions e.g. freqAlign/polarimetry against MS. Keep api.