[Coin-osi-devel] Next release of Osi

[Lou Hafer]

Folks,

	For the warm start methods, there's an issue that's come up with respect
to cbc.  Right now, there are three calls:

  * getEmptyWarmStart (returns an empty basis; a sort of specialised clone)
  * getWarmStart (return current warm start object)
  * setWarmStart (set a warm start in the OSI)

What's needed is a way to distinguish the actions `remove warm start info from
the OSI' and `request the OSI to synchronise its warm start information with the
solver'.

	The direct impetus comes because John and I have chosen conflicting
overloads for setWarmStart(0).  I've chosen this for `remove warm start info',
John uses it for `sync with solver'.

	My suggestion would be that setWarmStart(0) is naturally interpreted as
`remove warm start info' and we need a new call for `sync with solver'.  The
form of the solution isn't critical, but we need some way to distinguish the
two. Might as well tidy it up here.

	Depending on choices, we may need to revisit the default actions for
getWarmStart.  Right now, it returns an empty warm start if there's no valid
information.  Depending on the OSI implementation, this could violate the
const'ness of the call.

	I'll file this as a ticket on OSI.
	
						Lou